Skip to content

[Slides] Multimodal & video generation · Covers 2026-11-03 – 2026-11-1615 slides

Multimodal and Video Generation Brief · Nov 3–Nov 16, 2026

Generated with Divrite · Completed Nov 16, 2026

Get “Multimodal & video generation” by email · Free · No account needed · Unsubscribe anytime

Original prompt
Write the "Multimodal and Video Generation" hot-topic report covering 2026-11-03 to 2026-11-16. Unit of coverage: the model or product — track new releases in image, video, voice, music and 3D generation and compare quality, clip length, price and copyright policy. Sections: 1. Two-week highlights 2. New models and products 3. Quality and price comparison 4. Copyright and platform policy 5. Commercialization and the creator ecosystem 6. Trend assessment. Length: 12–15 slides. Key sources: official releases and samples, evaluation communities, platform policy announcements, creator communities. Cite every key number; leave out anything without a reliable source. Answer in English.

Opens Divrite with this prompt filled in — change it to your own topic before you start.

All 15 slides

  1. Slide 1: Multimodal and Video Generation Brief|Nov 3–Nov 16, 2026

    Slide 1

    Multimodal and Video Generation Brief|Nov 3–Nov 16, 2026

    Speaker notes

    Welcome to the Divrite Research Hot Topics brief on multimodal and video generation, covering November 3 to November 16, 2026. Across image, video, voice, music and 3D, we ask what shipped, what it costs, and which dates and lawsuits matter.

    Slide text

    IMAGE · VIDEO · VOICE · MUSIC · 3D

    Multimodal and Video Generation Brief

    15-second production clips at lower prices, a dated Google migration calendar, two new Suno lawsuits and money moving to distribution layers.

    PUBLISHER

    Divrite Research

    SERIES

    Hot Topics

    ISSUE

    Nov 3–Nov 16, 2026

  2. Slide 2: Contents

    Slide 2

    Contents

    Speaker notes

    This brief has five parts: an overview, new models, access and price, rights and money, and an outlook with dates to December 3. We begin with the overview.

    Slide text

    Contents | Five parts

    / 15

    PART

    Overview

    15-second clips, a dated calendar and two lawsuits

    PART

    Models

    Gemini Omni 1.2 Flash, Kling 4.0 Flash, Images 2.6, audio

    PART

    Access and price

    Dec 3 reroute, open-weight limits, Elo and cost

    PART

    Rights and money

    Suno suits, no new platform rule, Sand.ai $200M

    PART

    Outlook

    Trends, reader implications and dates to Dec 3

    Divrite Research · Multimodal and Video Generation Brief · Nov 3–Nov 16, 2026

  3. Slide 3: Overview|Price, dates and lawsuits outran new models

    Slide 3

    Overview|Price, dates and lawsuits outran new models

    Speaker notes

    Price, dates and lawsuits outran new models. Google made Gemini Omni 1.2 Flash generally available on November 5, with clips of up to 15 seconds at $0.07 per second of 720p video. Around it, the top five text-to-video models sit within 23 Elo, yet cost from $4.80 to $34.12 per minute, according to Artificial Analysis. Two Suno lawsuits plead about $202 million and $47.7 million, which are allegations, and Sand.ai raised $200 million, reported. So track dates, prices and filings, and choose by cost per clip.

    Slide text

    Overview

    Price, dates and lawsuits outran new models

    Source: Google; Kling AI; OpenAI; Artificial Analysis; Music Business Worldwide

    What moved in the window

    Nov 5: Gemini Omni 1.2 Flash generally available (GA)

    Nov 4: Kling 4.0 Flash open to everyone; OpenAI Images 2.6

    Nov 6: Presley rights holders and Eight Mile Style sue Suno (allegations)

    Nov 5: Sand.ai raises $200M (reported)

    Atlas: streamed at up to 15 fps (research preview)

    Not shipped: Midjourney V9 (estimate: early December)

    Key numbers

    15 s at $0.07 per second

    Omni 1.2 Flash at 720p; 30% below Omni 1.1

    23 Elo, $4.80 to $34.12

    top five text-to-video gap against price per minute

    $202M and $47.7M

    pleaded against Suno (allegations, not awards)

    $200M

    Sand.ai Series A+; valuation reported above $1B

    What it means: dated calendars, seller prices and court filings decide what builders can use more than any single model launch.

    Next step: track model IDs, prices and filings as dated items and choose by cost per clip.

    Where the numbers come from
    【本页依据】 ① Gemini Omni 1.2 Flash became generally available on Nov 5 with clips up to 15 s at $0.07 per second of 720p video, 30% below Omni 1.1. —— Gemini Omni 1.2 Flash: 15-second video and lower pricing · blog.google · https://blog.google/innovation-and-ai/technology/developers-tools/intr… Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blog · cloud.google.com · https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni… ② Kling 4.0 Flash was opened to everyone on Nov 4, and OpenAI introduced Images 2.6 on Nov 4. —— Kling AI on X: "Kling 4.0 Flash Is Now Fully Available! ⚡ Kling 4.0 Flash is now available to everyone on the Kling AI website and via API. You can also access it through MCP, CLI, and AI plugins, bringing greater flexibility to your video creation workflows. With Kling plugins for professional c… / X · x.com · https://x.com/Kling_ai/status/2117995299591319932;Introducing Images 2.6 · openai.com · https://openai.com/index/introducing-images-2-6 ③ World Labs streamed Atlas at up to 15 fps as a research preview on Nov 5. —— Frameboy: Atlas in Real-time | World Labs · worldlabs.ai · https://worldlabs.ai/blog/atlas-realtime ④ The top five text-to-video models on Artificial Analysis sit within 23 Elo (1175 to 1152) while prices run from $4.80 to $34.12 per minute. —— AA-Video-T2V v2.1 Leaderboard - Top AI Video Models | Artificial Analysis · artificialanalysis.ai · https://artificialanalysis.ai/video/leaderboard/text-to-video ⑤ The Presley rights holders and Eight Mile Style sued Suno on Nov 6 (allegations), pleading about $202M and $47.7M. —— Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an… Presley, Eminem's rights holders sue AI company Suno - Los Angeles Times · latimes.com · https://www.latimes.com/entertainment-arts/business/story/2026-11-09/e… ⑥ Sand.ai raised $200M in a Series A+ at a reported valuation above $1B. —— Sand.ai raised $200 million in a Series A+ round to build video generation models and tools for creators. CoStone Capital and Wangsu Science & Technology co-led the round. Sand.ai takes two routes… | Grishin Robotics · linkedin.com · https://www.linkedin.com/posts/grishinrobotics_sandai-raised-200-milli… Funding Has Sand.ai Raised? Full Round History | toflow.ai | toflow.ai · toflow.ai · https://toflow.ai/spotlight/sand-ai
  4. Slide 4: Video · Releases|Omni 1.2 Flash makes 15 s clips GA

    Slide 4

    Video · Releases|Omni 1.2 Flash makes 15 s clips GA

    Speaker notes

    Omni 1.2 Flash makes 15-second clips generally available. Google released it on November 5 on the Gemini API and Google Cloud, with multi-turn editing and up to six reference images; audio references are allowlisted. Kling AI said on November 4 that Kling 4.0 Flash is open to everyone, and World Labs showed Atlas at up to 15 frames per second as a research preview. At $0.07 per second, a 15-second clip costs $1.05. Test Omni 1.2 Flash against your current model at one resolution and compare cost per second.

    Slide text

    Video · Releases

    Omni 1.2 Flash makes 15 s clips GA

    Source: Google Gemini API blog and Google Cloud blog; Kling AI on X; Venice docs; World Labs

    What landed in the window

    Gemini Omni 1.2 Flash, Nov 5: GA on the Gemini API and Google Cloud; multi-turn editing, up to 6 reference images, SynthID watermark

    Audio references: allowlisted, not open to all

    Kling 4.0 Flash, Nov 4: open to everyone on website and API (Nov 5 per the Venice docs)

    Atlas, Nov 5: World Labs streams frames at up to 15 fps (research preview)

    Price: Omni 1.2 Flash is $0.07 per second at 720p, so 15 s costs $1.05

    Kling 4.0 Flash: from $0.1174 per second at a third-party host

    Context: Kling 4.0 launched Oct 15

    Max clip length, seconds (vendor claims)

    Read: window releases in blue; Kling 4.0 (Oct 15) is context

    What it means: at the production tier the contest is length and price, while world models stream frames instead of returning a fixed clip.

    Next step: test Omni 1.2 Flash against your current model at one resolution and compare cost per second.

    Where the numbers come from
    【本页依据】 ① Gemini Omni 1.2 Flash is GA on the Gemini API and Google Cloud with multi-turn editing, up to 6 reference images and a SynthID watermark. —— Gemini Omni 1.2 Flash: 15-second video and lower pricing · blog.google · https://blog.google/innovation-and-ai/technology/developers-tools/intr… AI Studio · ai.google.dev · https://ai.google.dev/gemini-api/docs/changelog;Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blog · cloud.google.com · https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni… ② Audio references for Omni 1.2 Flash are gated behind allowlisting. —— Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blog · cloud.google.com · https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni… ③ Kling AI said on Nov 4 that Kling 4.0 Flash is open to everyone on its website and API; the Venice docs list Nov 5 and from $0.1174 per second. —— Kling AI on X: "Kling 4.0 Flash Is Now Fully Available! ⚡ Kling 4.0 Flash is now available to everyone on the Kling AI website and via API. You can also access it through MCP, CLI, and AI plugins, bringing greater flexibility to your video creation workflows. With Kling plugins for professional c… / X · x.com · https://x.com/Kling_ai/status/2117995299591319932;Kling 4.0 Flash API: Pricing and Specs | Venice · docs.venice.ai · https://docs.venice.ai/models/kling-4-0-flash ④ World Labs streams Atlas frames at up to 15 fps (research preview), Nov 5. —— Frameboy: Atlas in Real-time | World Labs · worldlabs.ai · https://worldlabs.ai/blog/atlas-realtime ⑤ Omni 1.2 Flash is $0.07 per second at 720p, so a 15 s clip costs $1.05. —— Gemini Omni 1.2 Flash: 15-second video and lower pricing · blog.google · https://blog.google/innovation-and-ai/technology/developers-tools/intr… ⑥ Maximum clip length: Omni 1.1 Flash 10 s, Omni 1.2 Flash 15 s, Kling 4.0 Flash 20 s, Kling 4.0 30 s (vendor claims). —— Gemini Omni 1.2 Flash: 15-second video and lower pricing · blog.google · https://blog.google/innovation-and-ai/technology/developers-tools/intr… 4.0 Flash API: Pricing and Specs | Venice · docs.venice.ai · https://docs.venice.ai/models/kling-4-0-flash;Kling 4.0 API: access, pricing and code | Runway Dev · dev.runway.com · https://dev.runway.com/learn/kling-4-0-api
  5. Slide 5: Image · Audio|Faster images; audio moves into apps

    Slide 5

    Image · Audio|Faster images; audio moves into apps

    Speaker notes

    Images got faster and audio moved into apps. OpenAI introduced Images 2.6 on November 4, 50 percent faster, with GIFs and SVG export; ElevenLabs and NVIDIA expanded voice agents on November 11; and ElevenMusic can publish to SoundCloud. MiniMax's H3-Image has open weights, but Midjourney V9 is not released, and early December is only an estimate. No 3D model release is dated in the window. Pick image models by editing consistency and latency, and read licence terms before using open weights. OpenAI also says people create 4 billion images per week across its products, and MiniMax's 4096 by 4096 image in 10 to 20 seconds is a vendor claim.

    Slide text

    Image · Audio

    Faster images; audio moves into apps

    Source: OpenAI; CellCog; AI-0; ElevenLabs; TechCrunch

    What shipped outside video

    OpenAI Images 2.6, Nov 4: ChatGPT, Codex and API; GIFs and SVG export; Flare and Sunburst

    MiniMax H3-Image, Nov 10: open weights (Nov 11 per AI-0); 4096 x 4096 in 10 to 20 s (vendor claim)

    ElevenLabs and NVIDIA, Nov 11: voice-agent expansion with added compute

    ElevenMusic, Nov 5: publish straight to SoundCloud

    Midjourney V9: not released; early December is an estimate

    3D: no model release dated in the window

    Key numbers

    50% faster

    Images 2.6; Flare as little as six seconds at low quality

    4 billion

    images per week across OpenAI products

    25 million

    tracks created on ElevenLabs music models (company claim)

    15MW

    NVIDIA compute added by ElevenLabs

    What it means: image models compete on speed and editing, and audio products compete on where output can be published.

    Next step: pick image models by editing consistency and latency at your quality setting, and read licence terms before using open weights.

    Where the numbers come from
    【本页依据】 ① OpenAI introduced Images 2.6 on Nov 4 in ChatGPT, Codex and the API with GIFs, SVG export and Flare and Sunburst; it is 50% faster, and 4 billion images per week are created. —— Introducing Images 2.6 · openai.com · https://openai.com/index/introducing-images-2-6;ChatGPT Images 2.6 - OpenAI Deployment Safety Hub · deploymentsafety.openai.com · https://deploymentsafety.openai.com/chatgpt-images-2-6/chatgpt-images-… ② MiniMax announced open-weight H3-Image on Nov 10 (Nov 11 per AI-0), with a 4096 x 4096 image in 10 to 20 s as a vendor claim. —— MiniMax H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCog · cellcog.ai · https://cellcog.ai/blog/minimax-h3-image;AI video news November 2026: Kling 4.0, GPT Image 2.6 | AI-0 · ai-0.net · https://ai-0.net/blog/ai-video-image-news-november-2026 ③ ElevenLabs and NVIDIA expanded voice agents on Nov 11, with 15MW of added NVIDIA compute. —— ElevenLabs and NVIDIA expand voice agents for enterprise · elevenlabs.io · https://elevenlabs.io/blog/nvidia ④ ElevenMusic output can be published to SoundCloud (Nov 5); ElevenLabs says users have created over 25 million tracks. —— ElevenLabs now lets users publish AI-generated music directly on SoundCloud · tech.yahoo.com · https://tech.yahoo.com/audio/articles/elevenlabs-now-lets-users-publis… SoundCloud Deal: What We Know (Nov 2026) · mubert.com · https://mubert.com/blog/elevenlabs-soundcloud-elevenmusic-direct-publi… ⑤ Midjourney V9 is not released; early December is an estimate from secondary summaries of the Nov 4 office hours. —— What Is Midjourney v9? Specs, Changes, Roadmap · kie.ai · https://kie.ai/blog/what-is-midjourney-v9;🌹🦋 Timeless Visions AI Art Challenge Alerts 🌹🦋 | 🚀 MIDJOURNEY V9 IS GETTING CLOSER | Facebook · facebook.com · https://www.facebook.com/groups/1541771210401943/posts/1792145505364511
  6. Slide 6: Open weights|Check regional and revenue limits first

    Slide 6

    Open weights|Check regional and revenue limits first

    Speaker notes

    Check regional and revenue limits first. MiniMax's H3-Image, announced November 10, is the window's one open-weight release. According to CellCog and vdf.ai, MiniMax's licence excludes the US, EU, UK and South Korea, and revenue lines sit at $1 million for H3-Image, $10 million for LTX-2.5 and $20 million for MiniMax H3. Wan 3.0 has no downloadable weights. A download is not a clearance, so log each licence's territory and revenue threshold before deployment, or choose Apache or MIT models. The H3-Image weights are 33.1 billion parameters and 68.65 gigabytes, and these licence readings are secondary, so read the licence file itself.

    Slide text

    Open weights

    Check regional and revenue limits first

    Source: CellCog; vdf.ai; MarkTechPost

    What is open, and on what terms

    H3-Image, Nov 10: the window's one open-weight release; 33.1B parameters, 68.65 GB

    Territory: MiniMax licence excludes the US, EU, UK and South Korea

    Revenue lines: $1 million for H3-Image, $10 million for LTX-2.5, $20 million for MiniMax H3

    Permissive: Wan 2.2, Kandinsky 6.0 and MAGI-2-preview are Apache 2.0 or MIT

    Closed: Wan 3.0 has no downloadable weights

    Caveat: licence readings are secondary; read the licence file itself

    Revenue threshold in each licence, USD million

    Read: above each line, written permission or a paid licence

    What it means: a download is not a clearance: territory and revenue clauses decide commercial use.

    Next step: log each licence's territory and revenue threshold before deployment, or choose Apache or MIT models.

    Where the numbers come from
    【本页依据】 ① MiniMax H3-Image (Nov 10) has open weights of 33.1B parameters and 68.65 GB on Hugging Face. —— MiniMax H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCog · cellcog.ai · https://cellcog.ai/blog/minimax-h3-image ② The MiniMax licence excludes the US, EU, UK and South Korea. —— MiniMax H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCog · cellcog.ai · https://cellcog.ai/blog/minimax-h3-image;Best Local Video Generation Models 2026: VRAM and Licence · vdf.ai · https://vdf.ai/blog/best-local-video-generation-models ③ Revenue lines: $1 million for H3-Image, $10 million for LTX-2.5, $20 million for MiniMax H3. —— Same as ② ④ Wan 2.2, Kandinsky 6.0 and MAGI-2-preview are Apache 2.0 or MIT. —— Best Local Video Generation Models 2026: VRAM and Licence · vdf.ai · https://vdf.ai/blog/best-local-video-generation-models ⑤ Wan 3.0 has no downloadable weights. —— Best AI Video Generation Models in 2026: Elo, Price per Second, Open Weights - MarkTechPost · marktechpost.com · https://www.marktechpost.com/2026/11/13/best-ai-video-generation-model… Local Video Generation Models 2026: VRAM and Licence · vdf.ai · https://vdf.ai/blog/best-local-video-generation-models
  7. Slide 7: Access · Google|Dec 3 reroute; Nov 17 retirement floor

    Slide 7

    Access · Google|Dec 3 reroute; Nov 17 retirement floor

    Speaker notes

    Google's calendar is the one to watch. The Gemini API shut down its Veo 3.1 previews on October 22, and requests to Omni 1.1 Flash will route to Omni 1.2 Flash on December 3 at the new rates. On Google Cloud, veo-3.1-generate-001 lists a retirement date of November 17, 2026 or later, a floor and not an announced shutdown. Pin gemini-omni-1.2-flash now, test the new price, and read the lifecycle page before any Google Cloud cutover. Omni 1.2 Flash also supports up to six reference images and extension to 30 seconds total, with video output priced at $12.00 per million tokens.

    Slide text

    Access · Google

    Dec 3 reroute; Nov 17 retirement floor

    Source: Google AI for Developers release notes; Google Cloud docs

    Status on Nov 17

    Oct 22: Gemini API Veo 3.1 previews shut down (context)

    Dec 3: gemini-omni-1.1-flash routes to 1.2 at 1.2 rates

    Nov 17 or later: Google Cloud veo-3.1-generate-001 retirement date; a floor, not a shutdown

    Key numbers

    6 reference images, extension to 30 s total

    $12.00 per 1M tokens video output; paid-tier rate limits doubled

    Nov 17, 2025: release date of the Veo 3.1 GA models

    What it means

    Veo 3.1 now means different things on the Gemini API and on Google Cloud, and an Omni 1.1 call changes price on its own on Dec 3.

    Next step

    pin gemini-omni-1.2-flash now, test the new price, and read the lifecycle page before any Google Cloud cutover.

    Where the numbers come from
    【本页依据】 ① The Gemini API shut down its Veo 3.1 preview models on Oct 22 and names Gemini Omni Flash as the default video model. —— Google AI Studio · ai.google.dev · https://ai.google.dev/gemini-api/docs/veo;Video generation in the Gemini API  |  Google AI for Developers · ai.google.dev · https://ai.google.dev/gemini-api/docs/video ② Requests to gemini-omni-1.1-flash will route to gemini-omni-1.2-flash on Dec 3, 2026. —— Google AI Studio · ai.google.dev · https://ai.google.dev/gemini-api/docs/changelog ③ On Google Cloud, veo-3.1-generate-001 lists a retirement date of November 17, 2026 or later. —— Model versions and lifecycle  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentation · docs.cloud.google.com · https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/… 3.1  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentation · docs.cloud.google.com · https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/… ④ Omni 1.2 Flash supports up to 6 reference images, extension to 30 s total, doubled paid-tier rate limits and $12.00 per 1M video output tokens. —— Same as ② ⑤ The Veo 3.1 GA models were released Nov 17, 2025. —— Same as ③
  8. Slide 8: Quality · Elo|Top five sit within 23 Elo points

    Slide 8

    Quality · Elo|Top five sit within 23 Elo points

    Speaker notes

    The top five sit within 23 Elo points. On Artificial Analysis, whose page shows no as-of date, Wan 3.0 leads at 1175 and Omni 1.2 Flash is fifth at 1152, up 25 on Omni 1.1. Veo 3.1 sits at 963. On video editing, Wan 3.0 leads at 1192, and GPT Image 2.6 Sunburst leads images at 1259. Because rank ranges overlap and Elo says nothing about clip length, shortlist by price and task, then test on your own prompts. These are blind-vote ratings, and the video board scores clips with audio, so they are one view of quality, not a verdict.

    Slide text

    Quality · Elo

    Top five sit within 23 Elo points

    Source: Artificial Analysis (date not stated on the page)

    Artificial Analysis snapshot

    Pool: text to video with audio; the page shows no as-of date

    Top five: Wan 3.0, Utopai X, MiniMax H3 at 768p, Seedance 2.5, Omni 1.2 Flash

    Omni 1.2 Flash: 1152, up 25 on Omni 1.1 (1127); Veo 3.1 sits at 963

    Video editing: Wan 3.0 1192, Omni 1.2 Flash 1164

    Images: GPT Image 2.6 Sunburst 1259 against Flare 1228

    Caution: rank ranges overlap; Elo says nothing about clip length or editing control

    Text-to-video Elo (Artificial Analysis)

    Read: 23 points separate first from fifth (axis starts at 1000)

    What it means: places 1 to 5 are not cleanly separated, so price and workflow decide most choices.

    Next step: shortlist by price and task pool, then test on your own prompts.

    Where the numbers come from
    【本页依据】 ① Top five text-to-video with audio on Artificial Analysis: Wan 3.0 1175, Utopai X 1172, MiniMax H3 768p 1161, Seedance 2.5 1157, Omni 1.2 Flash 1152. —— AA-Video-T2V v2.1 Leaderboard - Top AI Video Models | Artificial Analysis · artificialanalysis.ai · https://artificialanalysis.ai/video/leaderboard/text-to-video ② Omni 1.2 Flash 1152 is up 25 on Omni 1.1 (1127); Veo 3.1 sits at 963. —— Same as ① ③ Video editing: Wan 3.0 1192, Omni 1.2 Flash 1164. —— AA-Video-Editing v1.1 Leaderboard - Top AI Video Models | Artificial Analysis · artificialanalysis.ai · https://artificialanalysis.ai/video/leaderboard/video-editing ④ Images: GPT Image 2.6 Sunburst 1259 against Flare 1228. —— AA-Image-T2I v2.1 Leaderboard - Top AI Image Models | Artificial Analysis · artificialanalysis.ai · https://artificialanalysis.ai/image/leaderboard/text-to-image
  9. Slide 9: Cost per clip|A 10 s 720p clip costs $0.70 to $4.00

    Slide 9

    Cost per clip|A 10 s 720p clip costs $0.70 to $4.00

    Speaker notes

    A ten-second 720p clip costs between $0.70 and $4.00. On Runway Dev's undated reseller list, Omni 1.2 Flash is 10 times $0.07, or $0.70, Wan 3.0 is $1.00, Gen-4.5 is $1.20, Kling 4.0 is $1.256, and Veo 3.1 with audio is $4.00. Kling 4.0 Flash is $0.094 at Runway but from $0.1174 at Venice. Compare cost per second at one resolution from one rate card, and re-check before the Kling 4.0 promotion ends on November 30. Veo 3.1 audio doubles its per-second price, from $0.20 to $0.40, and Runway plans work out between $0.096 and $0.230 per Gen-4.5 second.

    Slide text

    Cost per clip

    A 10 s 720p clip costs $0.70 to $4.00

    Source: Runway Dev; Runway; Venice; Artificial Analysis

    Seller price lists

    Source: Runway Dev pricing page, a reseller listing with no as-of date; 1 credit = $0.01

    Omni 1.2 Flash: 10 x $0.07 = $0.70; Wan 3.0 $1.00; Gen-4.5 $1.20; Kling 4.0 $1.256

    Audio: Veo 3.1 with audio is $4.00; audio doubles it ($0.40 against $0.20 per second)

    Kling 4.0 Flash: $0.094 at Runway, from $0.1174 at Venice; neither is Kling's own page

    Separate source: AA lists Omni 1.2 Flash at $6.26 per minute and Seedance 2.5 at $34.12

    Plans: Runway plans work out at $0.230, $0.149 and $0.096 per Gen-4.5 second

    Cost of a 10-second 720p clip, USD (Runway Dev list)

    Read: the same model costs different amounts by seller

    What it means: the same model costs different amounts by seller, and the cheapest is not the top-ranked.

    Next step: compare cost per second at one resolution and audio setting from one rate card; re-check before the Kling 4.0 promotion ends on Nov 30.

    Where the numbers come from
    【本页依据】 ① Runway Dev lists per-second prices: Omni 1.2 Flash $0.07, Wan 3.0 $0.10, Gen-4.5 $0.12, Kling 4.0 $0.1256, Veo 3.1 with audio $0.40. —— AI Video, Image and Audio API Pricing | Runway Dev · dev.runwayml.com · https://dev.runwayml.com/pricing ② Veo 3.1 audio doubles the per-second price ($0.40 against $0.20). —— Same as ① ③ Kling 4.0 Flash is $0.094 per output second at Runway and from $0.1174 at Venice. —— AI Video, Image and Audio API Pricing | Runway Dev · dev.runwayml.com · https://dev.runwayml.com/pricing;Kling 4.0 Flash API: Pricing and Specs | Venice · docs.venice.ai · https://docs.venice.ai/models/kling-4-0-flash ④ Artificial Analysis lists Omni 1.2 Flash at $6.26 per minute and Seedance 2.5 at $34.12. —— AA-Video-T2V v2.1 Leaderboard - Top AI Video Models | Artificial Analysis · artificialanalysis.ai · https://artificialanalysis.ai/video/leaderboard/text-to-video ⑤ Runway plans work out at $0.230, $0.149 and $0.096 per Gen-4.5 second. —— AI Image and Video Pricing from $12/month | Runway AI · runwayml.com · https://runwayml.com/pricing;AI Video, Image and Audio API Pricing | Runway Dev · dev.runwayml.com · https://dev.runwayml.com/pricing ⑥ Omni 1.2 Flash is $0.07 per second at 720p from Nov 5; LiteLLM lists Kling 4 on Runway at $0.1256 as of Nov 13. —— Gemini Omni 1.2 Flash: 15-second video and lower pricing · blog.google · https://blog.google/innovation-and-ai/technology/developers-tools/intr… 4 API Pricing (Nov 2026): $0.1256 per Second | LiteLLM · models.litellm.ai · https://models.litellm.ai/models/kling_4
  10. Slide 10: Litigation|Two Suno suits are claims, not rulings

    Slide 10

    Litigation|Two Suno suits are claims, not rulings

    Speaker notes

    The two Suno suits are claims, not rulings. On November 6 the Presley rights holders, over 859 compositions, and Eight Mile Style, over 269, each sued Suno in the Northern District of California; Suno denies both. The plaintiffs plead about $202 million and $47.7 million, according to Music Business Worldwide, while the Los Angeles Times says about $41 million for Eight Mile Style. Treat every figure as a claim, and watch December 1, when the response to Suno's motion to dismiss the Isbell class action is due.

    Slide text

    Litigation

    Two Suno suits are claims, not rulings

    Source: Music Business Worldwide; Los Angeles Times; Vibe; Courthouse News

    What was filed

    Nov 6: Presley rights holders v. Suno, N.D. Cal.: 859 compositions; ELVIS Act and copyright claims

    Nov 6: Eight Mile Style v. Suno, N.D. Cal.: 269 compositions; Eminem is not a plaintiff

    Suno: denies both; moved to dismiss the Isbell class action on Nov 3, response due Dec 1

    Arithmetic: 859 x $150,000 = $128.85M statutory; 269 x $150,000 = $40.35M

    Sources differ: Eight Mile Style about $41M (LA Times) against $47.7M total (MBW); we follow MBW

    Not verified: AFM suit over Suno and Udio licences (social post only)

    Amount pleaded, USD million (allegations)

    Read: ceilings the plaintiffs ask for, not awards

    What it means: pleaded exposure scales with the number of works listed, and the Presley case tests whether Tennessee's ELVIS Act reaches a general music generator.

    Next step: treat every figure as a claim and watch the Dec 1 response date.

    Where the numbers come from
    【本页依据】 ① On Nov 6 the Presley rights holders sued Suno in the Northern District of California over 859 compositions, including a claim under Tennessee's ELVIS Act. —— Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an… Presley, Eminem's rights holders sue AI company Suno - Los Angeles Times · latimes.com · https://www.latimes.com/entertainment-arts/business/story/2026-11-09/e… Presley rights holders sue AI music generator over licensing, copyright violations | Courthouse News Service · courthousenews.com · https://www.courthousenews.com/elvis-presley-rights-holders-sue-ai-mus… ② Eight Mile Style sued Suno on Nov 6 over 269 compositions; Eminem is not a plaintiff. —— Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an…'s Publisher Sues AI Music Generator Suno For $40M · vibe.com · https://www.vibe.com/music/music-news/eminems-publisher-40m-lawsuit-su… ③ Suno denies both suits and moved to dismiss the Isbell class action on Nov 3, with the response due Dec 1. —— Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an… rights holders and Eminem’s publisher sue Suno, testing the ELVIS Act · spokesman.com · https://www.spokesman.com/stories/2026/nov/09/elvis-rights-holders-and… ④ 859 x $150,000 = $128.85M statutory and about $202.47M pleaded; 269 x $150,000 = $40.35M and $47.7M pleaded. —— Same as ② ⑤ The Los Angeles Times says about $41M for Eight Mile Style against $47.7M in MBW. —— Elvis Presley, Eminem's rights holders sue AI company Suno - Los Angeles Times · latimes.com · https://www.latimes.com/entertainment-arts/business/story/2026-11-09/e… Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an… ⑥ A suit by the American Federation of Musicians over the Suno and Udio licences is described only in a social post. —— Instagram · instagram.com · https://www.instagram.com/reel/DfnFqUmS1T9
  11. Slide 11: Rules · Deals|No new platform rule; studios sign Adobe

    Slide 11

    Rules · Deals|No new platform rule; studios sign Adobe

    Speaker notes

    No new platform rule appeared, and studios signed with Adobe. The material shows no labelling change dated November 3 to 16, and the platforms' own pages were not collected. Adobe and Disney announced a multi-year alliance with Firefly Foundry models built on Disney IP, terms undisclosed. OpenAI says its images carry C2PA credentials and SynthID, but metadata can be stripped. Add AI disclosure and provenance checks to the publishing checklist, and keep source files with their credentials. The EU AI Act disclosure duty has been in effect since August 2, 2026, and Meta's advertiser clause took effect on October 30, both as context.

    Slide text

    Rules · Deals

    No new platform rule; studios sign Adobe

    Source: Adobe via PPC Land; OpenAI Help Center; Latent; Google Cloud blog

    What the material shows

    Platform rules: no labelling change dated Nov 3 to Nov 16; platform policy pages were not collected

    Adobe and Disney: multi-year alliance at Adobe MAX; Firefly Foundry models built on Disney IP; terms undisclosed

    Provenance: OpenAI images carry C2PA Content Credentials and SynthID; Omni 1.2 Flash outputs carry SynthID

    Warning: a screenshot or file conversion can strip metadata

    Context dates and figures

    Aug 2, 2026

    EU AI Act Article 50 disclosure duty in effect (context)

    EUR 15m or 3%

    maximum penalty by turnover under that duty

    Oct 30

    Meta Commercial Terms AI Features clause (context)

    $400M at $5.4B

    Suno's June round, per the Los Angeles Times (context)

    What it means: creators must disclose and mark synthetic output, but a credential proves neither truth nor cleared rights, and metadata can be stripped.

    Next step: add AI disclosure and provenance checks to the publishing checklist and keep source files with their credentials.

    Where the numbers come from
    【本页依据】 ① No platform labelling change dated Nov 3 to Nov 16 appears in the material. —— Report states that no platform-labelling change dated in the window appears in the material ② Adobe and Disney announced a multi-year alliance with Firefly Foundry models built on Disney IP; terms undisclosed. —— Adobe gains marketing tie-ins with 5 Disney units in multi-year AI alliance · ppc.land · https://ppc.land/adobe-gains-marketing-tie-ins-with-5-disney-units-in-… ③ OpenAI images carry C2PA Content Credentials and SynthID; a screenshot or file conversion can strip metadata. —— Provenance signals in content processed by OpenAI tools | OpenAI Help Center · help.openai.com · https://help.openai.com/en/articles/8912793-provenance-signals-in-cont… ④ Omni 1.2 Flash outputs carry an imperceptible SynthID watermark. —— Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blog · cloud.google.com · https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni… ⑤ EU AI Act Article 50 disclosure duty in effect since Aug 2, 2026, with penalties up to EUR 15m or 3% of turnover; Meta Commercial Terms AI Features clause Oct 30. —— Is AI-generated advertising allowed? The 2026 rules | Latent · thesillyhumans.com · https://thesillyhumans.com/latent-ai-advertising-rules-2026;AI Labels Explained: What They Mean for UGC and Ad Performance - Sideshift Blog · sideshift.app · https://sideshift.app/blog/ai-labels ⑥ Suno raised $400M at $5.4B in June. —— Elvis rights holders and Eminem’s publisher sue Suno, testing the ELVIS Act · spokesman.com · https://www.spokesman.com/stories/2026/nov/09/elvis-rights-holders-and…
  12. Slide 12: Funding|Sand.ai raised $200M; valuations are claims

    Slide 12

    Funding|Sand.ai raised $200M; valuations are claims

    Speaker notes

    Sand.ai raised $200 million, but valuations are claims. The Series A+ was co-led by CoStone Capital and Wangsu Science and Technology, with a valuation reported above $1 billion, from an investor post and a sales-intelligence page rather than a filing. Sand.ai says its VidMuse app passed $10 million in annual recurring revenue within two months. Tracker valuations cluster at $4.0 to $5.4 billion. Treat vendor revenue and valuation figures as unverified, and separate announced terms from estimates. fal's October round is context, and sources disagree on its size, at $1.3 billion or $1.65 billion.

    Slide text

    Funding

    Sand.ai raised $200M; valuations are claims

    Source: Grishin Robotics post; toflow.ai; New Market Pitch; Silicon Valley Investclub

    Money in the window

    Sand.ai, Nov 5: $200M Series A+, co-led by CoStone Capital and Wangsu; valuation reported above $1B

    Evidence: an investor post and a sales-intelligence page, not a filing

    Company claim: VidMuse passed $10M annual recurring revenue within two months of launch

    fal (context): $1.3B or $1.65B at $16B; sources disagree

    Tracker: separates 24 valuations backed by statements or terms from 52 estimates

    Tracker valuations, USD billion (New Market Pitch, Nov 14)

    Read: top valuations cluster at $4.0B to $5.4B

    What it means: money is landing on workflow and product companies as well as model labs, and every growth number here is self-reported.

    Next step: treat vendor revenue and valuation figures as unverified, and separate announced terms from estimates.

    Where the numbers come from
    【本页依据】 ① Sand.ai raised a $200M Series A+ co-led by CoStone Capital and Wangsu Science & Technology; valuation reported above $1B. —— Sand.ai raised $200 million in a Series A+ round to build video generation models and tools for creators. CoStone Capital and Wangsu Science & Technology co-led the round. Sand.ai takes two routes… | Grishin Robotics · linkedin.com · https://www.linkedin.com/posts/grishinrobotics_sandai-raised-200-milli… Funding Has Sand.ai Raised? Full Round History | toflow.ai | toflow.ai · toflow.ai · https://toflow.ai/spotlight/sand-ai ② The Sand.ai figures come from an investor post and a sales-intelligence page. —— Same as ① ③ Sand.ai says VidMuse passed $10M annual recurring revenue within two months of launch. —— Sand.ai raised $200 million in a Series A+ round to build video generation models and tools for creators. CoStone Capital and Wangsu Science & Technology co-led the round. Sand.ai takes two routes… | Grishin Robotics · linkedin.com · https://www.linkedin.com/posts/grishinrobotics_sandai-raised-200-milli… ④ fal raised at a $16B valuation in October; the round is reported as $1.3B or $1.65B. —— 𝐟𝐚𝐥 𝐑𝐚𝐢𝐬𝐞𝐬 $𝟏.𝟑 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐚𝐭 𝐚 $𝟏𝟔 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐕𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧 fal is a startup that hosts and runs cutting-edge AI video, image, and audio models. Designed to empower developers and creators with high-performance infrastructure, the company provides scalable APIs and fast tools that eliminate deployment bottlenecks, making it easy to integrate state-of-the-art generative AI into real-time applications. fal has raised $1.3 billion in its latest funding round, bringing the company’s valuation to $16 billion. The round was led by Bessemer Venture Partners, with participation from Nvidia, Sound Ventures, and T. Rowe Price. fal, headquartered in San Francisco, California, United States, was founded in 2021 by Burkay Gur and Gorkem Yurtseven. In 2025, the company joined the unicorn club. “This year, we at fal pushed the limits of what’s possible, of what’s imaginable, in generative media. We post-trained already outstanding models to be even better and faster, in order to test what happens when quality converges with speed. We also cemented generative media’s impact in every major industry, expanding the very definition of what a creative layer can enable,” wrote the company in its post. | Silicon Valley Investclub · facebook.com · https://www.facebook.com/siliconvalleyinvestclub/videos/%F0%9D%90%9F%F… · instagram.com · https://www.instagram.com/p/DfSxIDsH00R ⑤ Tracker valuations: Higgsfield $5.4B, Runway $5.3B, Decart, Luma AI and Synthesia $4.0B each; 24 valuations are backed by statements or terms and 52 are estimates. —— What are the most valuable AI video generation startups? – New Market Pitch · newmarketpitch.com · https://newmarketpitch.com/blogs/news/ai-video-generation-top-startups…
  13. Slide 13: Distribution|Adobe and Runway resell rival models

    Slide 13

    Distribution|Adobe and Runway resell rival models

    Speaker notes

    Adobe and Runway resell rival models. At Adobe MAX on November 10, Premiere's Generative Media tool offered Firefly plus Google, Kling, Runway and Luma models, and Runway Dev sells the Kling 4.0 API self-serve. LiteLLM lists Kling 4 at $0.1256 per second on Runway as of November 13. Pricing power moves to whoever owns the creator's workflow, so prefer surfaces that let you switch models, and keep prompts, references and edits outside any one model. Kling 4.0 on Runway shows how one maker's model reaches creators through several sellers, and Adobe also announced $1 million in direct grants to ten filmmakers.

    Slide text

    Distribution

    Adobe and Runway resell rival models

    Source: Adobe; Runway Dev; LiteLLM; TechCrunch; SaaS CRM Review

    Who sells what

    Adobe MAX, Nov 10: Premiere tool offers Firefly, Google, Kling, Runway and Luma models

    Runway Dev: sells the Kling 4.0 API self-serve, with a Model Router

    Music: ElevenMusic to SoundCloud; licensed UMG and WMG music in Premiere mobile

    Key numbers

    $0.1256 per second Kling 4 on Runway (Nov 13)

    $39 and $8.40 a month Higgsfield Plus and Hailuo AI Standard (Oct 2026, context)

    $1M Adobe Film and TV Fund grants to 10 filmmakers

    What it means

    pricing power moves to whoever owns the creator's workflow, and a model can be repriced, rerouted or left to decay.

    Next step

    prefer surfaces that let you switch models, and keep prompts, references and edits outside any one model.

    Where the numbers come from
    【本页依据】 ① At Adobe MAX (Nov 10) Premiere's Generative Media tool offers Firefly plus Google, Kling, Runway and Luma models. —— Explore new ways to create across Creative Cloud and Adobe Firefly while advancing your craft with more precision, choice and control · blog.adobe.com · https://blog.adobe.com/en/publish/2026/11/10/explore-new-ways-create-a… ② Google's latest imaging and video models are due to come to Firefly, Photoshop, Premiere and Express. —— Adobe and Google Deepen Partnership to Bring Leading AI Models and Integrated Workflows to Creatives and Enterprises · news.adobe.com · https://news.adobe.com/news/2026/11/adobe-and-google-deepen-partnership ③ Runway Dev sells the Kling 4.0 API self-serve with a Model Router. —— Kling 4.0 API: access, pricing and code | Runway Dev · dev.runway.com · https://dev.runway.com/learn/kling-4-0-api ④ ElevenMusic output can be published to SoundCloud, and Premiere mobile adds licensed UMG and WMG music. —— ElevenLabs now lets users publish AI-generated music directly on SoundCloud · tech.yahoo.com · https://tech.yahoo.com/audio/articles/elevenlabs-now-lets-users-publis… mobile adds music from UMG and WMG, helping creators find the perfect track for every story · blog.adobe.com · https://blog.adobe.com/en/publish/2026/11/03/premiere-mobile-adds-musi… ⑤ LiteLLM lists Kling 4 at $0.1256 per second on Runway as of Nov 13. —— Kling 4 API Pricing (Nov 2026): $0.1256 per Second | LiteLLM · models.litellm.ai · https://models.litellm.ai/models/kling_4 ⑥ Higgsfield Plus is $39 per month and Hailuo AI Standard $8.40 per month on an Oct 2026 list; Adobe Film and TV Fund gave $1M to 10 filmmakers. —— 10 Google Flow Alternatives in 2026: Veo, Credits & Cost · saascrmreview.com · https://saascrmreview.com/google-flow-alternatives;Adobe backs the next generation of filmmakers with $1 million in direct funding and new industry Advisory Board · blog.adobe.com · https://blog.adobe.com/en/publish/2026/11/09/adobe-backs-next-generati…
  14. Slide 14: Outlook · Dates|Watch Nov 30, Dec 1 and Dec 3 next

    Slide 14

    Outlook · Dates|Watch Nov 30, Dec 1 and Dec 3 next

    Speaker notes

    Watch November 30, December 1 and December 3. Longer, cheaper clips and aggregators are strong trends, licensing alongside litigation and open weights with strings are mixed, and world models are weak. On November 30 Runway Dev's Kling 4.0 promotion ends, on December 1 the Isbell response is due, and on December 3 Omni 1.1 Flash reroutes to 1.2. Midjourney V9 is an early-December estimate. Developers pin model IDs, creators compare cost per second, rights holders watch December 1. No ruling on training is scheduled in the material, so the dated risks ahead are retirements, reroutes, promotions and court deadlines.

    Slide text

    Outlook · Dates

    Watch Nov 30, Dec 1 and Dec 3 next

    Source: Google Cloud docs; Gemini API release notes; Runway Dev; Music Business Worldwide

    Trends and confidence

    Longer, cheaper clips: strong

    Aggregators beside model makers: strong

    Licensing alongside litigation: mixed

    Open weights with strings attached: mixed

    Single-model dependence risk: mixed

    World models: weak; one research demo

    Dates to watch

    Nov 17

    Vertex veo-3.1-generate-001 retirement floor; H3-Image 4K promotion ends

    Nov 30

    Runway Dev Kling 4.0 promotion ends

    Dec 1

    plaintiffs' response in the Isbell class action

    Dec 3

    Omni 1.1 Flash reroute to 1.2; Midjourney V9 is an early-December estimate

    What it means: the dated risks ahead are retirements, reroutes, promotions and court deadlines; no ruling on training is scheduled in the material.

    Next step: developers pin model IDs, creators compare cost per second, rights holders watch the Dec 1 filing.

    Where the numbers come from
    【本页依据】 ① Longer, cheaper clips are a strong trend: Omni 1.2 Flash 15 s at $0.07 per second and Kling 4.0 Flash 3 to 20 s. —— Gemini Omni 1.2 Flash: 15-second video and lower pricing · blog.google · https://blog.google/innovation-and-ai/technology/developers-tools/intr… 4.0 Flash API: Pricing and Specs | Venice · docs.venice.ai · https://docs.venice.ai/models/kling-4-0-flash;AA-Video-T2V v2.1 Leaderboard - Top AI Video Models | Artificial Analysis · artificialanalysis.ai · https://artificialanalysis.ai/video/leaderboard/text-to-video ② Aggregators beside model makers: Premiere offers several makers' models and Runway Dev sells Kling 4.0. —— Explore new ways to create across Creative Cloud and Adobe Firefly while advancing your craft with more precision, choice and control · blog.adobe.com · https://blog.adobe.com/en/publish/2026/11/10/explore-new-ways-create-a… 4.0 API: access, pricing and code | Runway Dev · dev.runway.com · https://dev.runway.com/learn/kling-4-0-api;Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blog · cloud.google.com · https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni… ③ Licensing alongside litigation, and open weights with strings attached, are mixed. —— Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an… mobile adds music from UMG and WMG, helping creators find the perfect track for every story · blog.adobe.com · https://blog.adobe.com/en/publish/2026/11/03/premiere-mobile-adds-musi… now lets users publish AI-generated music directly on SoundCloud · tech.yahoo.com · https://tech.yahoo.com/audio/articles/elevenlabs-now-lets-users-publis… H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCog · cellcog.ai · https://cellcog.ai/blog/minimax-h3-image;Best Local Video Generation Models 2026: VRAM and Licence · vdf.ai · https://vdf.ai/blog/best-local-video-generation-models;Best AI Video Generation Models in 2026: Elo, Price per Second, Open Weights - MarkTechPost · marktechpost.com · https://www.marktechpost.com/2026/11/13/best-ai-video-generation-model… ④ Single-model dependence: about 60 Grok Imagine team members were cut (reported) and Veo 3.1 previews left the Gemini API. —— Grok Imagine Loses Most of Its Video Team: Keep Your Film Portable · screenweaver.ai · https://www.screenweaver.ai/blog/grok-imagine-team-cut-ai-filmmakers;G… AI Studio · ai.google.dev · https://ai.google.dev/gemini-api/docs/veo ⑤ Nov 17 Vertex veo-3.1-generate-001 retirement floor and H3-Image half-price 4K promotion end; Nov 30 Runway Dev Kling 4.0 promotion ends. —— Model versions and lifecycle  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentation · docs.cloud.google.com · https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/… H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCog · cellcog.ai · https://cellcog.ai/blog/minimax-h3-image;AI Video, Image and Audio API Pricing | Runway Dev · dev.runwayml.com · https://dev.runwayml.com/pricing ⑥ Dec 1 plaintiffs' response in the Isbell class action; Dec 3 Omni 1.1 Flash reroute; Midjourney V9 an early-December estimate. —— Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide · musicbusinessworldwide.com · https://www.musicbusinessworldwide.com/elvis-presleys-rightsholders-an… AI Studio · ai.google.dev · https://ai.google.dev/gemini-api/docs/changelog;What Is Midjourney v9? Specs, Changes, Roadmap · kie.ai · https://kie.ai/blog/what-is-midjourney-v9
  15. Slide 15: Dated facts decide what you can build on

    Slide 15

    Dated facts decide what you can build on

    Speaker notes

    Closing. Dated facts decide what you can build on. Models ship fast; the dates, prices and licences around them decide what creators can use. Thank you.

    Slide text

    Closing

    Models ship fast; the dates, prices and licences

    around them decide what creators can use.

    Thank you

    2026-11-17 · Multimodal and Video Generation Brief · Nov 3–Nov 16, 2026

Sources (156)

Links go to the original pages Divrite found and cited while researching.

  1. 01Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blogcloud.google.com
  2. 02Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blogcloud.google.com
  3. 03Gemini Omni 1.2 Flash now GA: cost-effective videocloud.google.com
  4. 04Gemini Omni 1.2 Flash: 15-second video and lower pricingblog.google
  5. 05Gemini Omni 1.2 Flash: 15-second video and lower pricingblog.google
  6. 06Gemini Omni 1.2 Flash: 15-second video and lower pricingdeepmind.google
  7. 07GitHub - useapi/kling-api: Kling API examples by useapi.net: Kling 4.0, 3.0 and Omni video (text-to-video, image-to-video with keyframes, multi-shot, motion control, native sound), lip-sync avatars, KOLORS images and a hosted MCP server (45 tools), on your Kling AI subscription, with no Kling developer account. · GitHubgithub.com
  8. 08Kling 4.0 Flash API: Pricing and Specs | Venicedocs.venice.ai
  9. 09Kling 4 API Pricing (Nov 2026): $0.1256 per Second | LiteLLMmodels.litellm.ai
  10. 10Kling AI on X: "Kling 4.0 Flash Is Now Fully Available! ⚡ Kling 4.0 Flash is now available to everyone on the Kling AI website and via API. You can also access it through MCP, CLI, and AI plugins, bringing greater flexibility to your video creation workflows. With Kling plugins for professional c… / Xx.com
  11. 11Kling 4.0 API: Released Oct 15, 2026, 30s Video | UnificAllyunifically.com
  12. 12Kling 4.0 API: access, pricing and code | Runway Devdev.runway.com
Show 144 more sources
  1. 13MiniMax H3 (Hailuo 3) AI Video: 2K Clips with Native Soundocmaker.ai
  2. 14MiniMax H3-Image: a 4K AI image in under 10 seconds, with up to 16 reference images in one prompt. Released on November 10 and already live in Hailuo AI, in the MiniMax API and as open weights on Hugging Face. • Speed: a full 4096 × 4096 frame in under 10 seconds, native sizes up to 21:9, no upscaler. • Realism: MiniMax is going after the overly smooth, artificial AI look. • References: your character, your location, your props, called by number in the prompt. One model generates and edits. In the official samples, a grey 3D render becomes a photoreal shot, a painted map becomes a landscape, and the same fox stays the same fox across three frames. That's what a keyframe needs. But a great frame is still just a frame. ScreenWeaver reads your script, breaks every scene down shot by shot, and keeps your characters consistent across the whole storyboard, so you know which frames to generate. Send this to the friend who still waits a minute for every 4K render. Comment STORYBOARD and I'll send you the link, or go straight to https://www.screenweaver.ai/start Sources: MiniMax blog, VP Land (Nov 10). #aiimage #minimax #hailuoai #aifilmmaking #keyframes #aiart #storyboard #aifilm #filmmaking #screenweaver | Frank Houbre AIfacebook.com
  3. 15MiniMax H3 Image Generator Online: Native 4K | Flux1 AIflux1.ai
  4. 16Alibaba Next-Gen Video Model: What's Confirmed So Far | each::labseachlabs.ai
  5. 17Hailuo 2.3: Arena rank 17, $0.41 per 5 s clipgradually.ai
  6. 18MiniMax Hailuo 2.3 API: Pricing, Sample Video, Docs - Wiro AIwiro.ai
  7. 19Best AI Video Generators 2026: Kling, Veo, Runway | M Studiomstudio.ai
  8. 20AI video news November 2026: Kling 4.0, GPT Image 2.6 | AI-0ai-0.net
  9. 21Grok Imagine Loses Most of Its Video Team: Keep Your Film Portablescreenweaver.ai
  10. 227 Best Pika Alternatives for AI Video in 2026higgsfield.ai
  11. 23Apps — GrabStackgrabstack.com
  12. 24Best AI Video Models Sept 2026: Ranked by Price per Secondteamday.ai
  13. 25World Labs Acquires SceniX | World Labsworldlabs.ai
  14. 26Frameboy: Atlas in Real-time | World Labsworldlabs.ai
  15. 27Frameboy: Atlas in Real-time | World Labsworldlabs.ai
  16. 28Atlas | World Labsworldlabs.ai
  17. 29MCP server · Atlasdocs.atlas.worldlabs.ai
  18. 30Introducing Images 2.6openai.com
  19. 31ChatGPT Images 2.6 System Card - OpenAI Deployment Safety Hubdeploymentsafety.openai.com
  20. 32ChatGPT Images 2.6 - OpenAI Deployment Safety Hubdeploymentsafety.openai.com
  21. 33OpenAI | Research & Deploymentopenai.com
  22. 34Images, upgraded—introducing ChatGPT Images 2.6 - Announcements - OpenAI Developer Communitycommunity.openai.com
  23. 35Responses API with gpt-image-2 gives `Complexity 'low' is not supported` after gpt-image-2.6 launch (with temporary hack) - Bugs - OpenAI Developer Communitycommunity.openai.com
  24. 36What Is Midjourney v9? Specs, Changes, Roadmapkie.ai
  25. 37Free AI Art Image Video Music Prompt Share Group Artgoai.com | 🚀 MIDJOURNEY V9 IS GETTING CLOSER | Facebookfacebook.com
  26. 38🌹🦋 Timeless Visions AI Art Challenge Alerts 🌹🦋 | 🚀 MIDJOURNEY V9 IS GETTING CLOSER | Facebookfacebook.com
  27. 39Midjourney Images & Insights. Share, Discuss Your Best Creations, Prompts. | 🚀 MIDJOURNEY V9 IS GETTING CLOSER | Facebookfacebook.com
  28. 40Midjourney V9 Is About 30 Days Out. Here’s What I’m Doing Before It Lands.medium.com
  29. 41Creativity AI #101: Midjourney V9 is weeks away, ChatGPT makes GIFs, and Claude Fable 5.2 arrivesgeekycuriosity.substack.com
  30. 42Explore new ways to create across Creative Cloud and Adobe Firefly while advancing your craft with more precision, choice and controlblog.adobe.com
  31. 43Entdeckt neue Möglichkeiten zum Gestalten mit Creative Cloud und Adobe Firefly, während ihr eure Fähigkeiten mit mehr Präzision, Auswahl und Kontrolle weiterentwickeltblog.adobe.com
  32. 44Adobe MAX 2026: Eine neue Ära der Kreativität – mit den passenden Tools für Kreative überallnews.adobe.com
  33. 45Adobe and Google Deepen Partnership to Bring Leading AI Models and Integrated Workflows to Creatives and Enterprisesnews.adobe.com
  34. 46creative-cloud-release-max-2026.pdfnews.adobe.com
  35. 47Adobe Firefly entwickelt sich zum KI-Kreativstudio weiter und macht Adobes bewährte Kreativwerkzeuge für alle Kreativen zugänglichnews.adobe.com
  36. 48ElevenLabs and NVIDIA expand voice agents for enterpriseelevenlabs.io
  37. 49Best Voice LLMs for Voice Agents in 2026: Ranked | Covalcoval.ai
  38. 50Eleven v4 & v4 Turbo: What’s New and Live Demos of AI Agentselevenlabs.io
  39. 511.2K views · 19 reactions | We’re expanding our collaboration with ElevenLabs to bring voice agents to enterprises and governments worldwide. With NVIDIA Nemotron models now in ElevenAgents, organizations have more choice in how they build and customize voice agents. NVIDIA’s Kari Briski joined ElevenLabs’ Alex Holt today to discuss the news and bringing enterprise AI into production. Read the news: elevenlabs.io/blog/nvidia | NVIDIAfacebook.com
  40. 52ElevenLabs’ new app combines meeting note-taking and dictation for teamstechcrunch.com
  41. 53Suno vs Udio vs ElevenLabs & More: Which AI Music Generator Sounds Best in 2026?youtube.com
  42. 54Butch Warner Musicfacebook.com
  43. 55The 10 Best AI Music Generators in 2026gradually.ai
  44. 56ElevenLabs SoundCloud Integration Launches for AI Musicemergent.sh
  45. 57ElevenLabs SoundCloud Deal: What We Know (Nov 2026)mubert.com
  46. 58ElevenLabs now lets users publish AI-generated music directly on SoundCloudtech.yahoo.com
  47. 59Gemini API changelog - Google AIai.google.dev
  48. 60Google AI Studioai.google.dev
  49. 61Google AI Studioai.google.dev
  50. 62Google AI Studioai.google.dev
  51. 63Veo 3.1  |  Gemini API  |  Google AI for Developersai.google.dev
  52. 64Video generation in the Gemini API  |  Google AI for Developersai.google.dev
  53. 65Model versions and lifecycle  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentationdocs.cloud.google.com
  54. 66Veo 3.1  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentationdocs.cloud.google.com
  55. 67Vertex AI release notes  |  Google Cloud Documentationdocs.cloud.google.com
  56. 68Deployments and endpoints  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentationdocs.cloud.google.com
  57. 69Extend videos  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentationdocs.cloud.google.com
  58. 70Generate videos from references  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentationdocs.cloud.google.com
  59. 71Changelog | OpenAI APIplatform.openai.com
  60. 72Changelog | OpenAI APIplatform.openai.com
  61. 73Deprecations | OpenAI APIplatform.openai.com
  62. 74Deprecations | OpenAI APIplatform.openai.com
  63. 75Deprecations | OpenAI APIplatform.openai.com
  64. 76Changelog | OpenAI APIplatform.openai.com
  65. 77Best AI Video Generation Models in 2026: Elo, Price per Second, Open Weights - MarkTechPostmarktechpost.com
  66. 78Best Local Video Generation Models 2026: VRAM and Licencevdf.ai
  67. 79Instagraminstagram.com
  68. 80MiniMax H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCogcellcog.ai
  69. 81Model Release Velocity | Superpower Dailysuperpowerdaily.com
  70. 82Hugging Face News, Products, Models & Company Updates | Superpower Dailysuperpowerdaily.com
  71. 83AI 3D Model Generators for Game Development (2026) | Wavectwavect.io
  72. 84November 2026 AI Model Updates: Every Launch, Access Change, and Price Move - Local AI Zonelocal-ai-zone.github.io
  73. 85Open Source AI Models: 3 Run Paths, 48% Index [2026]shattered.io
  74. 86Best Open Source AI Models in 2026 (Ranked)felloai.com
  75. 87AA-Video-T2V v2.1 Leaderboard - Top AI Video Models | Artificial Analysisartificialanalysis.ai
  76. 88AA-Video-T2V v2.1 排行榜 - 顶级 AI 视频模型 | Artificial Analysisartificialanalysis.ai
  77. 89Ranking AA-Video-T2V v2.1 - Principais modelos de vídeo com IA | Artificial Analysisartificialanalysis.ai
  78. 90AA-Video-Editing v1.1 Leaderboard - Top AI Video Models | Artificial Analysisartificialanalysis.ai
  79. 91AA-Video-Editing v1.1 Leaderboard - Top AI Video Models | Artificial Analysisartificialanalysis.ai
  80. 92Ranking de AA-Video-Editing v1.1 - Mejores modelos de video con IA | Artificial Analysisartificialanalysis.ai
  81. 93AA-Image-T2I v2.1 Leaderboard - Top AI Image Models | Artificial Analysisartificialanalysis.ai
  82. 94AA-Image-Editing v2.1 Leaderboard - Top AI Image Models | Artificial Analysisartificialanalysis.ai
  83. 95Ranking de AA-Image-T2I v2.1 - Mejores modelos de IA para imagen | Artificial Analysisartificialanalysis.ai
  84. 96Image Model Comparisonsartificialanalysis.ai
  85. 97AI Model & API Providers Analysis | Artificial Analysisartificialanalysis.ai
  86. 98Image Arena - Top AI Image Models | Artificial Analysisartificialanalysis.ai
  87. 99Video Arena Is Live on Web - Arena.aiarena.ai
  88. 100How to Generate Videos with Video Arena | Arena Help Centerhelp.arena.ai
  89. 101How to see AI Rankings in Arena Leaderboards | Arena Help Centerhelp.arena.ai
  90. 102Arena Blogarena.ai
  91. 103Google AI Studioai.google.dev
  92. 104Gemini Developer API pricing - Gemini API | Google AI Studioai.google.dev
  93. 105Gemini Omni Flash - Gemini API | Google AI Studioai.google.dev
  94. 106Gemini Omni Flash - Gemini API | Google AI Studioai.google.dev
  95. 107Gemini Omni Flash - Gemini API | Google AI Studioai.google.dev
  96. 108AI Image and Video Pricing from $12/month | Runway AIrunwayml.com
  97. 109AI Video, Image and Audio API Pricing | Runway Devdev.runwayml.com
  98. 110API Pricing & Costs | Runway Devdocs.dev.runwayml.com
  99. 111Standard plan details – Runwayhelp.runwayml.com
  100. 112How do credits work? – Runwayhelp.runwayml.com
  101. 113Creating with Gen-4.5 – Runwayhelp.runwayml.com
  102. 114Elvis rights holders and Eminem’s publisher sue Suno, testing the ELVIS Actspokesman.com
  103. 115Elvis Presley, Eminem's rights holders sue AI company Suno - Los Angeles Timeslatimes.com
  104. 116Eminem Is Suing Suno For $41,000,000 - Here's What It Actually Meansyoutube.com
  105. 117Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwidemusicbusinessworldwide.com
  106. 118Elvis Presley rights holders sue AI music generator over licensing, copyright violations | Courthouse News Servicecourthousenews.com
  107. 119Eminem's Publisher Sues AI Music Generator Suno For $40Mvibe.com
  108. 120Instagraminstagram.com
  109. 121Adobe gains marketing tie-ins with 5 Disney units in multi-year AI allianceppc.land
  110. 122How to Tell If a Video Is AI Generated: 7 Signs (2026)revid.ai
  111. 123AI Content Labels Explained (2026): Meta, YouTube and TikTok Disclosure Rulesyoutube.com
  112. 124When Should You Label Videos as 'AI'?kapwing.com
  113. 125AI Labels Explained: What They Mean for UGC and Ad Performance - Sideshift Blogsideshift.app
  114. 126Is AI-generated advertising allowed? The 2026 rules | Latentthesillyhumans.com
  115. 127Social Media Updates You Must Know in 2026 (November Edition)blog.gainapp.com
  116. 128Googleprimo.ai
  117. 129Do AI Images Have Invisible Watermarks? | Watermark.wswatermark.ws
  118. 130AI Watermarks Explained: SynthID, C2PA and How Undetectr Removes Themyoutube.com
  119. 131India Mandates 24-Hour Deepfake Takedowns and C2PA Labels | AIFODaf.net
  120. 132Provenance signals in content processed by OpenAI tools | OpenAI Help Centerhelp.openai.com
  121. 133Instagraminstagram.com
  122. 134#ai #videogeneration #funding | Startup Researcherlinkedin.com
  123. 135Instagraminstagram.com
  124. 136Sand.ai raised $200 million in a Series A+ round to build video generation models and tools for creators. CoStone Capital and Wangsu Science & Technology co-led the round. Sand.ai takes two routes… | Grishin Roboticslinkedin.com
  125. 137What Funding Has Sand.ai Raised? Full Round History | toflow.ai | toflow.aitoflow.ai
  126. 138Ventureburnfacebook.com
  127. 139What are the most valuable AI video generation startups? – New Market Pitchnewmarketpitch.com
  128. 140𝐟𝐚𝐥 𝐑𝐚𝐢𝐬𝐞𝐬 $𝟏.𝟑 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐚𝐭 𝐚 $𝟏𝟔 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐕𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧 fal is a startup that hosts and runs cutting-edge AI video, image, and audio models. Designed to empower developers and creators with high-performance infrastructure, the company provides scalable APIs and fast tools that eliminate deployment bottlenecks, making it easy to integrate state-of-the-art generative AI into real-time applications. fal has raised $1.3 billion in its latest funding round, bringing the company’s valuation to $16 billion. The round was led by Bessemer Venture Partners, with participation from Nvidia, Sound Ventures, and T. Rowe Price. fal, headquartered in San Francisco, California, United States, was founded in 2021 by Burkay Gur and Gorkem Yurtseven. In 2025, the company joined the unicorn club. “This year, we at fal pushed the limits of what’s possible, of what’s imaginable, in generative media. We post-trained already outstanding models to be even better and faster, in order to test what happens when quality converges with speed. We also cemented generative media’s impact in every major industry, expanding the very definition of what a creative layer can enable,” wrote the company in its post. | Silicon Valley Investclubfacebook.com
  129. 141Instagraminstagram.com
  130. 142Video Startups funded by Y Combinator (YC) 2026 | Y Combinatorycombinator.com
  131. 143Top Companies in AI Video Generator (Oct, 2026) - Tracxntracxn.com
  132. 144Instagraminstagram.com
  133. 14510 Google Flow Alternatives in 2026: Veo, Credits & Costsaascrmreview.com
  134. 146Cross-platform guides: AI guides | Dreamdrivedreamdrive.ai
  135. 147Higgsfield AI Review 2026: Pricing & Best Alternatives - Global Think Labglobalthinklab.com
  136. 148Premiere mobile adds music from UMG and WMG, helping creators find the perfect track for every storyblog.adobe.com
  137. 149高い精度、豊富な選択肢、柔軟なコントロール性を持つ新しい制作アプローチをAdobe Creative CloudとAdobe Fireflyで体験blog.adobe.com
  138. 150Adobe backs the next generation of filmmakers with $1 million in direct funding and new industry Advisory Boardblog.adobe.com
  139. 151Adobe for Startupsの提供を開始blog.adobe.com
  140. 152YouTube AI Content Policy: Disclosure & Monetizationlayer3labs.io
  141. 153YouTube AI Content Monetization Policy 2026: What's Allowedtubechef.ai
  142. 154YouTube Automation (2026): Tools, Rules and 3 Workflowsrevid.ai
  143. 155YouTube Tightens Monetization Rules for AI-Generated Contenttechrepublic.com
  144. 156Best AI Video Generator for Long YouTube Videos (2026) | HeyGenheygen.com
Contents
  1. 011. Key takeaways from the two weeks
  2. 022. New video and world models
  3. 033. New image, voice, music and 3D models
  4. 044. Access, retirements and open weights
  5. 055. Quality and price comparison
  6. 066. Copyright and platform policy
  7. 077. Commercialization and the creator ecosystem
  8. 088. Trend assessment and what to watch next
  9. 099. References

1. Key takeaways from the two weeks

Between Nov 3 and Nov 16, 2026 the multimodal market shipped one flagship video release, one major image release and a lot of distribution news, while the dated risks sit in Google's model calendar and in two new music lawsuits. Five conclusions follow from Chapters 2 to 7.

Document cover: abstract light background for the multimodal and video generation brief.
Document cover: abstract light background for the multimodal and video generation brief.

1.1 Five conclusions

1. At the production tier the contest is length and price: Google made 15-second clips GA at $0.07 per second, and the research frontier moved to real-time worlds. Gemini Omni 1.2 Flash became generally available on Nov 5 with native clips of up to 15 seconds (from 10) and a 30% price cut to $0.07 per second of 720p video. Kling AI said on Nov 4 that Kling 4.0 Flash was open to everyone on its website and API, and World Labs streamed its Atlas world model at up to 15 frames per second in a browser demo (research preview). On images, OpenAI's Images 2.6 (Nov 4) is 50% faster and adds GIFs; Midjourney V9 did not ship and is a reported estimate for around early December. [Chapter 2] [Chapter 3]

2. For builders, the calendar matters more than the model list. Gemini API requests to gemini-omni-1.1-flash will be rerouted to Omni 1.2 Flash on Dec 3, 2026; the Vertex veo-3.1-generate-001 models list a retirement date of "November 17, 2026 or later", which is a floor, not an announced shutdown; and the one in-window open-weight release, MiniMax's H3-Image, carries a licence that excludes the US, EU, UK and South Korea. [Chapter 4]

3. Quality is bunched, so price separates the leaders. On Artificial Analysis' text-to-video board (no as-of date on the page) the top five sit within 23 Elo points (1175 to 1152) while normalized prices run from $4.80 to $34.12 per minute; GPT Image 2.6 Sunburst leads the next image model by 31 points (1259 against 1228). Resellers disagree even on one model: Kling 4.0 Flash is $0.094 per second at Runway and "from $0.1174" at Venice. [Chapter 5]

4. Rights holders keep litigating; the window's cases are allegations against one music generator, and no platform rule changed. On Nov 6 the Elvis Presley rights holders and Eight Mile Style (Eminem's publisher) each sued Suno in the Northern District of California, pleading just over $202 million and about $47.7 million respectively; Suno denies both. Suno's Nov 3 motion to dismiss a separate class action has a response due Dec 1. No platform-labelling change dated Nov 3 to Nov 16 appears in the material. [Chapter 6]

5. Money and distribution are concentrating on the layer that owns the creator's workflow. Sand.ai, a Beijing video-model startup, raised $200 million at a reported valuation above $1 billion; at Adobe MAX (Nov 10) Premiere's Generative Media tool offers models from Google, Kling, Runway and Luma; and ElevenMusic output can be published directly to SoundCloud. [Chapter 7]

The window moved less on raw capability than on access, price and rules: a 15-second clip at $0.07 per second, a dated Google migration calendar, and two lawsuits that test voice and likeness law. Distribution surfaces, not single models, are where creators now choose.

1.2 What changed at a glance

Model / product Change Date Status Source
Premiere mobile Licensed UMG and WMG music added Nov 3 Announced Adobe
Images 2.6 (OpenAI) 50% faster, GIFs, API Flare and Sunburst Nov 4 GA OpenAI
Kling 4.0 Flash Opened to all on website and API Nov 4 (Kling post) or Nov 5 (Venice docs) API available Kling AI; Venice
Gemini Omni 1.2 Flash 15-second clips, $0.07 per second at 720p Nov 5 GA Google
Atlas "Frameboy" (World Labs) Real-time demo at up to 15 fps Nov 5 Research preview World Labs
ElevenMusic and SoundCloud Direct publishing Nov 5 Announced TechCrunch via Yahoo
Sand.ai $200M Series A+ Nov 5 (toflow.ai) Reported Grishin Robotics post; toflow.ai
Presley rights holders and Eight Mile Style v. Suno Two copyright complaints Nov 6 Allegations; pending MBW
Adobe MAX Generative Media tool; Google partnership Nov 10 Announced Adobe
MiniMax H3-Image Open weights, 4K claim Nov 10 (CellCog) or Nov 11 (AI-0) Open weights; vendor claims CellCog; AI-0
ElevenLabs and NVIDIA Voice-agent expansion, 15MW added compute Nov 11 Announced ElevenLabs
Grok Imagine team About 60 people cut, product still live Reported Nov 13 Reported ScreenWeaver citing The Information
Vertex veo-3.1-generate-001 Retirement date listed Nov 17, 2026 or later Upcoming Google Cloud docs

2. New video and world models

Chapter 1 summarises the window; on the video side the two weeks from Nov 3 to Nov 16 hold one flagship release, one price-led sibling and one real-time research demo, and everything else below is context.

2.1 Release timeline

One production release, one cheaper sibling: Google made Gemini Omni 1.2 Flash generally available on Nov 5 with native clips of up to 15 seconds and a 30% price cut to $0.07 per second of 720p video, and Kling AI said on Nov 4 that Kling 4.0 Flash was "available to everyone" on its website and API.

Specs below are vendor claims unless stated; arena ratings and per-second prices are in Chapter 5.

Model Maker Date Access Max clip Resolution Native audio Status
Gemini Omni 1.2 Flash Google DeepMind / Google Cloud Nov 5 (window) Gemini API, Google Cloud, Gemini for business, Google Vids 15 s (3 to 15 s in 1 s steps) Drafts at 360p or 720p, upscale to 1080p or 4K Dialogue, lip-sync, multi-speaker; audio references allowlisted GA
Kling 4.0 Flash Kuaishou (Kling AI) Nov 4 per Kling's X post; Nov 5 per the Venice API docs (window) Kling website, API, MCP, CLI, plug-ins 20 s (3 to 20 s) 720p Always generated API available
Kling 4.0 Kuaishou (Kling AI) Oct 15 (context) App; API via Runway Dev and others 30 s 720p, 1080p Stereo Context
MiniMax H3 (Hailuo 3) MiniMax Before the window (context) Hosted by third parties, for example OCMaker 5 to 15 s on that host 480p, 720p, 2K on that host Yes Context; host figures
Atlas "Frameboy" demo World Labs Nov 5 (window) Browser demo, research preview Streamed, not clips Up to 15 frames per second Not applicable Research preview

According to the AI-0 roundup (Nov 14), no dated new release from Veo, Sora, Seedance, Wan, Runway, Grok Imagine, Midjourney or Luma fell in the Oct 15 to Nov 14 stretch beyond the items above; this is a secondary source and a negative finding, so we treat it as "none found" rather than "none happened". Pika H3 Flex (Oct 29, app Nov 2) is context.

2.2 What is genuinely new

Gemini Omni 1.2 Flash is the substantive release. Google's own post lists five changes from Omni 1.1: clips up to 15 seconds (from 10), a price 30% lower, stronger adherence to reference images and short motion clips of up to three seconds, multi-turn conversational editing of a finished clip, and more natural dialogue with clearer lip-sync. The Google Cloud post adds that it is "production-ready" with Provisioned Throughput, that outputs carry an imperceptible SynthID watermark, and that audio references are gated behind allowlisting. Named launch customers (Artlist, FLORA, invideo, Krea, Picsart, Photoroom and Runway) are vendor-quoted testimonials, not independent tests. The cut is the story for builders: $0.07 per second at 720p means a 15-second clip costs $1.05 (15 x $0.07), before any upscale to 1080p or 4K.

Kling 4.0 Flash is a lower-cost variant of Kling 4.0, not a new generation. The Venice API docs, a third-party host, list it as released Nov 5, 720p only, 3 to 20 seconds, audio always generated, from $0.1174 per second (a 10-second clip is $1.17). Kling's own post is dated Nov 4 and also lists plug-ins for Photoshop, Premiere Pro and After Effects. The one-day gap between the two sources is probably a time-zone effect, but the material does not say; we follow the vendor's Nov 4 post for the announcement and the host's Nov 5 for the API listing. The full Kling 4.0 (up to 30 seconds, 720p or 1080p, stereo audio) predates the window. Runway Dev now sells the Kling 4.0 API self-serve, an example of a competitor reselling a model (see Chapter 7).

2.3 World models versus clip generators

World Labs' Nov 5 research preview shows the other direction of travel. Atlas, which the company introduced about two months earlier, is an autoregressive world model that generates frames one at a time with camera control; "Frameboy" streams frames from cloud GPUs to a browser at up to 15 frames per second while the user steers the camera, and a streaming overlay was added to the Viewfinder scene-building app. The Atlas page lists a developer API, Blender integration and splat-to-mesh conversion, and the docs list an MCP server that runs Atlas tasks from Claude and other MCP clients using the project's developer API credits. This is a research preview and a developer tool for scene exploration, not a clip generator; there is no clip length to compare, and the two categories should not share a ranking.

One negative signal sits at the other end of the market. The Information reported on Nov 13 that about 60 people, more than half of the Grok Imagine team, were let go over one week; the product remains online and no roadmap was announced. We read the report through a secondary summary (it is paywalled), so the number is "reported", and SpaceX had not commented at that time. Read it as a retention risk for creators who build on one model, which is the reason aggregators that route across several models are gaining ground (Chapter 7).

Takeaway: the window rewards price and length at the production tier (15 seconds at $0.07 per second), while the research frontier moves from clips to interactive worlds; neither has an independent quality ranking dated inside the window beyond Chapter 5's arena snapshot.

3. New image, voice, music and 3D models

Chapter 2 covered video and world models; outside video the window is led by one major image release (OpenAI), one open-weight image release (MiniMax), a tool-and-partner push at Adobe MAX, and a run of audio-side distribution news, with no dated 3D model release in this chapter's material.

3.1 Image models

Image models are getting faster and gaining motion: OpenAI's Images 2.6 (Nov 4) generates images 50% faster than before, adds animated GIFs and, in the API, SVG export, markup and a new complexity parameter. The same week MiniMax announced H3-Image with open weights (Nov 10).

Model Maker Date Access Licence Headline capability Status
Images 2.6 (GPT-Image-2.6 Flare and Sunburst) OpenAI Nov 4 (window) ChatGPT, Codex, API Closed 50% faster; Flare "as little as six seconds" at low quality; GIFs; SVG export; image markup in the API GA
H3-Image MiniMax Nov 10 per CellCog; Nov 11 per AI-0 (window) Hailuo AI, MiniMax API, Hugging Face weights Open weights under MiniMax's own licence (terms in Chapter 4) 4096 x 4096 (about 16.8 million pixels) in 10 to 20 seconds, up to 16 reference images; vendor claim Open weights, per secondary sources
Seedream 5.5 ByteDance Seed Nov 9 per AI-0 (window) Hosted Not stated More natural skin, sharper text, transparent backgrounds Reported by a secondary roundup
Midjourney V9 Midjourney Not released Not available Closed Native full-resolution base model plus new upscaler, in internal testing Upcoming, estimate

OpenAI's own numbers are the only primary ones here. The company says people now create 4 billion images per week across ChatGPT, Codex and the API, "more than half a billion images created per day". The system card dated 2026-11-04 describes the safety stack as unchanged from Images 2.5 plus extra safeguards for GIF creation. A developer thread on the OpenAI community forum complains that low and medium quality settings bill substantially more output tokens than gpt-image-2.5; that is a user claim, not OpenAI's statement, so cost comparisons are deferred to Chapter 5.

Midjourney V9 did not ship. Secondary summaries of the Nov 4 Midjourney office hours say a functional full-resolution base model exists internally, that the team is deciding between a full launch and an early preview, and that a release "within approximately 30 days of November 4" was the target; Midjourney said V9 images "could cost double". That is a reported estimate, not an announcement, and nothing in the material shows a release by Nov 16. The same sources list native video with sound, real-time generation and 3D camera controls as roadmap items, not shipped features.

3.2 Voice and speech

ElevenLabs' window was about enterprise voice infrastructure rather than a new voice model. On Nov 11 the company and NVIDIA said NVIDIA's Nemotron 3.5 Super VL is now available in ElevenAgents, that ElevenLabs is adding 15MW of NVIDIA Blackwell Ultra and Vera Rubin compute, and that it powers a sovereign voice agent for Poland's national e-Health platform. On Nov 13 TechCrunch reported "ElevenAI", an experimental meeting-notes and dictation app launched at the company's 11/11 Summit, still in preview with early-access sign-up. No new text-to-speech model release dated inside the window appears in the material, so none is claimed.

3.3 Music

Item Maker Date Access What changed Status
ElevenMusic to SoundCloud publishing ElevenLabs, SoundCloud Nov 5 (window) ElevenMusic users publish directly to SoundCloud; SoundCloud Artist Pro users get three months of ElevenMusic Pro Distribution, not a new model Announced, available per TechCrunch
Licensed music in Premiere mobile Adobe with UMG and WMG Nov 3 (window) Free Premiere mobile app Licensed music clips from two major labels Announced

TechCrunch notes that ElevenLabs says users have created over 25 million tracks with its models (a company claim) and that streaming platforms "seem more inclined toward labeling and helping users filter out AI-generated music". The pattern is the same in both rows: music generation is moving into distribution and editing surfaces that already have audiences, with licensed catalogue (UMG, WMG) on one side and generated catalogue on the other. No new Suno or Udio model release dated in the window is supported by the material (the Suno news is legal, see Chapter 6).

3.4 Adobe MAX: creative tools with partner models

Adobe MAX 2026 (Nov 10) put generative audio and video into the editing timeline. Premiere's new Generative Media tool offers Generate Video, Generate Sound Effects, Generate Soundscape and Generate Music in the timeline, and the user can choose Adobe Firefly or partner models from Google, Kling, Runway and Luma. Adobe and Google also announced a wider partnership: Google's latest imaging and video models will come to Firefly, Photoshop, Premiere and Express, and Adobe and Google DeepMind will collaborate on image editing starting with Photoshop (a statement of intent, no date given). Pricing and distribution consequences are in Chapter 7. There was no 3D model release in this chapter's material; Atlas, a world model with 3D outputs, is in Chapter 2.

4. Access, retirements and open weights

Chapters 2 and 3 described what shipped; this chapter covers what builders can actually call, what is being switched off, and which weights can be downloaded and under what terms.

4.1 Deprecation and retirement calendar

The Google calendar decides who has to move: the Gemini API's Veo 3.1 preview models were shut down on Oct 22, 2026 with Gemini Omni Flash named as the replacement, the Vertex-side Veo 3.1 GA models list a retirement date of "November 17, 2026 or later", and Omni 1.1 Flash traffic will be rerouted to Omni 1.2 Flash on Dec 3, 2026.

Gemini API and Google Cloud model chain: Veo 3.1 previews, Omni 1.1 Flash reroute on Dec 3, Omni 1.2 Flash GA and the Vertex Veo 3.1 retirement floor.
Gemini API and Google Cloud model chain: Veo 3.1 previews, Omni 1.1 Flash reroute on Dec 3, Omni 1.2 Flash GA and the Vertex Veo 3.1 retirement floor.
Product Action Date Replacement Status on 2026-11-17
Veo 3.1 Preview (veo-3.1-generate-preview, -fast-, -lite-) in the Gemini API Shut down Oct 22, 2026 Gemini Omni Flash (gemini-omni-1.2-flash) or the GA models on the Gemini Enterprise Agent Platform In effect; context, before the window
veo-3.1-generate-001, veo-3.1-fast-generate-001 (Gemini Enterprise Agent Platform) Retirement date listed "November 17, 2026 or later"; released Nov 17, 2025 Not named on the lifecycle page Listed as GA, retirement date is the run date or later; no earlier date shown
gemini-omni-1.1-flash (Gemini API) Automatic routing to Omni 1.2 Flash at 1.2 rates Dec 3, 2026 gemini-omni-1.2-flash Upcoming; announced Nov 5
Gemini Omni 1.2 Flash General availability Nov 5, 2026 Not applicable In effect

Two readings matter. First, the Vertex footnote says listed retirement dates refer to support in the Gemini Enterprise Agent Platform and that models "may remain accessible through the Gemini API" after those dates; the Gemini API Veo page, however, says its preview models are already shut down, so "Veo 3.1" now means different things on the two surfaces. Second, "November 17, 2026 or later" is a floor, not a promise of removal on that day; the material shows no replacement notice, so the page should be checked directly before any production cutover. The Gemini API video guide now tells developers to use Gemini Omni Flash as the default video model and keeps Veo as the second option. The collected OpenAI changelog and deprecations pages show no image-model retirement dated in the window; the earlier Sora shutdown is context only.

4.2 API versus app availability

The release notes make the Gemini side concrete: Omni 1.2 Flash supports first-and-last-frame interpolation of 3 to 10 seconds, per-turn extension of 3 to 10 seconds up to 30 seconds total, up to 6 reference images and doubled paid-tier rate limits, with the video output price at $12.00 per 1M tokens. That is the API picture behind the consumer-facing 15-second headline in Chapter 2. For Kling 4.0 Flash, World Labs Atlas and OpenAI's Images 2.6 the access routes (API, MCP, plug-ins) are in Chapters 2 and 3; none is gated beyond ordinary account sign-up in the material, except Google's allowlisting of audio references (Chapter 2). Chapter 5 holds the prices.

4.3 Open weights and licences

The material gives one open-weight release inside the window and several earlier releases that decide what is usable commercially. All licence descriptions below come from secondary sources that read the licence files; the licence text itself should be read before use.

Model Maker Released Licence per source Catch Status
H3-Image (image, 33.1B parameters, 68.65 GB) MiniMax Nov 10 (window) MiniMax's own licence Excludes the US, EU, UK and South Korea; firms above $1 million revenue need written permission Open weights, CellCog reading
MiniMax H3 (video, H3-Base 768p ceiling) MiniMax Aug 3 weights (context) MiniMax H3 Community License Same four territories excluded; written authorisation above $20 million yearly revenue Open weights, context
LTX-2.5 Lightricks Aug 2026 (context) LTX-2.x Community License (dated Aug 11, amended Oct 29) Free below $10 million annual revenue; watermark and provenance features must stay Open weights, context
Wan 2.2 family, Kandinsky 6.0, MAGI-2-preview Alibaba, Kandinsky Lab, Sand AI 2025 to Sep 2026 (context) Apache 2.0 or MIT Commercial use allowed Open weights, context
Wan 3.0 Alibaba Aug 24 (context) None No downloadable weights; the Wan-AI Hugging Face organisation had none for Wan 2.5 or later on Nov 4 Hosted only

The practical split is territory and revenue-capped licences (MiniMax, LTX) versus permissive licences (Wan 2.2, Kandinsky, MAGI-2) that are older or less capable. According to MarkTechPost's Nov 13 ranking, which uses Artificial Analysis Elo, the top-ranked text-to-video model (Wan 3.0) is closed, while the open-weight MiniMax H3 sits in the top three; arena detail is in Chapter 5. For 3D, a secondary guide checked on Nov 4 lists Microsoft's TRELLIS.2 (4B image-to-3D, MIT code and model licence, NVIDIA GPU with at least 24GB VRAM) as the main self-hosted option and notes that Tripo separates its H3.1 and P2 models; no 3D open-weight release dated inside the window appears in the material.

4.4 What to do before the next date

  • Gemini API users on Omni 1.1 Flash: rerouting to 1.2 happens on Dec 3 at 1.2 rates, so cost per second changes by itself; pin gemini-omni-1.2-flash now if you want to test the new price first.
  • Teams still on Vertex Veo 3.1 GA models: treat Nov 17, 2026 as the earliest retirement, not the actual one, and read the lifecycle page weekly.
  • Anyone using open weights commercially in the US, EU, UK or South Korea: MiniMax's licences exclude those territories; choose Apache or MIT models instead.

5. Quality and price comparison

Chapters 2 to 4 said what shipped and what can be called; this chapter asks what the new models score and cost, using one source per table, and it separates ratings, prices and vendor specs.

5.1 Leaderboard snapshot

Quality is bunched, so price decides: on Artificial Analysis' text-to-video board the top five sit within 23 Elo points (1175 to 1152) while their API prices run from $4.80 to $34.12 per minute; on images, GPT Image 2.6 Sunburst leads the runner-up by 31 points.

Artificial Analysis (AA) ranks by Elo from blind pairwise votes. None of its pages shows an as-of date, so every rating below is "date not stated on the page; crawled for this issue". The Portuguese copy of the video board shows 7,651 votes for Wan 3.0 against 7,787 on the English page, so the pages were captured at different times. arena.ai is not tabulated: the collected pages are a Jan 21, 2026 launch post and a help article with no ratings.

AA-Video-T2V v2.1 (text to video, with audio), source: ref_87

Rank (range) Model Elo 95% interval Votes Released
1 (1-2) Wan 3.0 1175 1162-1188 7,787 Aug 2026
2 (1-3) Utopai X 1172 1158-1186 7,171 Sep 2026
3 (2-6) MiniMax H3 (768p) 1161 1148-1174 7,948 Jul 2026
4 (3-8) Dreamina Seedance 2.5 1157 1144-1170 7,907 Jul 2026
5 (3-9) Gemini Omni Flash 1.2 1152 1138-1166 5,354 Nov 2026
13 (10-14) Gemini Omni Flash 1.1 1127 1114-1140 8,253 Aug 2026
25 (25-26) Veo 3.1 963 949-977 5,048 Jan 2026

Omni 1.2 Flash, the window's flagship (Chapter 2), enters at rank 5 with the fewest votes of the top five, 25 Elo above its predecessor (1152 - 1127); its rank range of 3 to 9 means the board cannot separate it from FLUX 3 Max or HeyGen Video 1. On AA's video-editing board it is second at 1164, behind Wan 3.0 at 1192. According to MarkTechPost (Nov 13, a secondary ranking built on AA), Seedance 2.5 leads image to video at 1161 and Omni 1.2 Flash scores 1086 there; the same article's text-to-video figures differ from the crawled page by 1 to 2 points (Omni 1154 against 1152), which is ordinary drift.

AA image boards, source: ref_93 (text to image) and ref_94 (editing)

Model Text-to-image Elo (rank) Editing Elo (rank)
GPT Image 2.6 Sunburst (max) 1259 (1) 1210 (1)
GPT Image 2.6 Flare (max) 1228 (2) 1206 (2)
GPT Image 2.5 Sunburst (max) 1227 (3) 1201 (3)
Nano Banana 2.1 1176 (6) 1152 (9)
Grok Imagine Image 2.0 1166 (8) 1117 (23)

OpenAI's Images 2.6 (Nov 4, Chapter 3) beats Images 2.5 Sunburst by 32 points on text to image (1259 - 1227), outside the 1244-1274 interval, but by only 9 on editing (1210 - 1201), where all three leaders share ranks 1 to 4. MiniMax H3-Image (Chapter 3) is 13th on editing at 1138, and Seedream 5.5 is 7th at 1159.

5.2 Price table

Prices below come from Runway Dev's pricing page, one reseller listing many makers (1 credit = $0.01); the page shows no as-of date, so it is "crawled for this issue", and the Kling 4.0 promotion runs through Nov 30, 2026.

Model (maker) Unit Resolution Price
Gemini Omni 1.2 Flash (Google) per second 720p / 1080p / 4K $0.07 / $0.10 / $0.21
Gemini Omni 1.1 Flash (Google) per second 720p $0.10
Kling 4.0 Flash (Kling) per output second, no input video 720p $0.094
Kling 4.0 (Kling) per second, 20% off 720p / 1080p $0.1256 (list $0.157) / $0.188 (list $0.235)
Gen-4.5 (Runway) per second not listed $0.12
Wan 3.0 (Alibaba) per second 720p / 1080p $0.10 / $0.20
Seedance 2.5 (ByteDance) per second 720p / 1080p $0.30 / $0.68
Veo 3.1 (Google), with audio per second 720p and 1080p $0.40

Arithmetic for a 10-second 720p clip on this list: Omni 1.2 Flash 10 x $0.07 = $0.70; Wan 3.0 10 x $0.10 = $1.00; Gen-4.5 10 x $0.12 = $1.20; Kling 4.0 10 x $0.1256 = $1.256; Veo 3.1 with audio 10 x $0.40 = $4.00. Google's own figure agrees with its resale: $0.07 per second at 720p from Nov 5, billed in the Gemini API at $12.00 per 1M video-output tokens; the material does not give tokens per second, so we do not convert. LiteLLM's Nov 13 listing of Kling 4 at $0.1256 matches Runway.

Kling 4.0 Flash disagrees across hosts: Runway lists $0.094 per output second, Venice "from $0.1174" per second at 720p with audio. The Venice figure is 25% higher (0.1174 / 0.094 = 1.249). Neither is Kling's own price page, which the material lacks, so we show both and follow neither as a vendor price.

AA's own normalized video prices (1080p where offered, per minute) are a separate source: Wan 3.0 $12.00 (= $0.20 per second), MiniMax H3 768p $4.80 (= $0.08), Omni 1.2 Flash $6.26 (= $0.1043), Seedance 2.5 $34.12 (= $0.569). So Wan 3.0 costs 1.9 times as much as Omni 1.2 Flash (12.00 / 6.26) for 23 more Elo points. For images AA lists $210.72 per 1,000 for all four OpenAI "max" variants (= $0.2107 per image), against $33.60 for Nano Banana 2.1 (= $0.0336), 6.3 times less for 83 fewer Elo points (1259 - 1176). MiniMax's example 1K H3-Image cost is $0.02179 (= $21.79 per 1,000), under AA's $26.54 per 1,000, and MiniMax says its examples are not fixed prices.

Runway's consumer plans (annual billing) convert as follows at 12 credits per second of Gen-4.5: Standard $12 / 625 credits x 12 = $0.230 per second; Pro $28 / 2,250 x 12 = $0.149; Max $76 / 9,500 x 12 = $0.096, against $0.12 on the API. The page shows no date.

5.3 Clip length and audio

All specs here are vendor or reseller statements, not test results.

Model Clip length Resolution Audio Source
Gemini Omni 1.2 Flash 3-15 s; extension to 30 s total 360p to 4K, 24 FPS Native Gemini API docs
Kling 4.0 Flash 3-20 s 720p Always generated Venice docs
Gen-4.5 2-10 s 720p, 24 or 25 FPS Not stated Runway help

Audio is also a price lever. Veo 3.1 costs $0.40 per second with audio and $0.20 without (audio doubles it); Kling 3.0 Pro is $0.17 with and $0.12 without, 42% more. AA's text-to-video board scores clips with audio; its silent board has a different leader, HeyGen Video 1 at 1146, with Wan 3.0 fifth at 1136.

5.4 What the comparison does not show

Elo measures blind preference on short prompts, not clip length, consistency or editing control, and a 3-to-9 rank range means ties. AA prices a minute at 1080p, while Google's headline is 720p, so $0.07 and $0.1043 are different products. Credit and token billing hide rounding, minimums and failed generations (per CellCog quoting MiniMax's guide, charges apply to successful images even if the client disconnects). No creator head-to-heads or arena.ai ratings were collected, so editing workflow, prompt adherence in a real project and brand-safe output remain untested here.

6. Copyright and platform policy

Chapter 5 compared quality and price; the legal side of the window is dominated by one defendant, Suno, which was sued twice in one day, while the platform-rule material contains no new rule dated inside the window.

6.1 Litigation and settlements

Two new complaints against Suno, filed Nov 6, are allegations, not findings: the Presley rights holders plead about $202 million across their claims and Eight Mile Style about $47.7 million; nothing has been decided and Suno denies both.

Both complaints were filed on November 6, 2026 in the US District Court for the Northern District of California, by the same legal team led by attorney Richard Busch.

Case Court Plaintiffs Core claims (allegations) Amount pleaded Status on 2026-11-17
Presley rights holders v. Suno N.D. Cal., filed Nov 6 (window) Elvis Presley Enterprises, ABG EPE IP (both controlled by Authentic Brands Group), Raleigh Music Publishing, Abilene Music and others Copyright infringement over 859 compositions; Tennessee ELVIS Act; right of publicity; circumvention of access controls; removal of credits Statutory damages $128.85 million, plus $2.1475 million and $21.475 million, plus "believed to exceed $50 million" for voice and likeness: just over $202 million Complaint filed; Suno denies; no ruling
Eight Mile Style v. Suno N.D. Cal., filed Nov 6 (window) Eight Mile Style and Martin Affiliated (Eminem is not a plaintiff) Copyright infringement over 269 compositions; altered or removed copyright metadata Just over $47.7 million pleaded in total Complaint filed; Suno denies; no ruling

The arithmetic matches MBW's count. 859 works x $150,000 (the statutory maximum for willful infringement) = $128.85 million. Adding $2.1475 million, $21.475 million and one $50 million claim gives $202.47 million, "just over $202 million". For Eight Mile Style, 269 x $150,000 = $40.35 million, which is the figure Vibe reports as "more than $40.35 million"; the Los Angeles Times says "about $41 million"; MBW's $47.7 million is the sum of all pleaded claims. I follow MBW's $47.7 million as the total and $40.35 million as the statutory component, because MBW states how the number is built. Each figure is a ceiling the plaintiffs ask for, not an award.

The Presley complaint is one of the first tests of Tennessee's ELVIS Act (in effect since 2024), which targets unauthorized commercial use of a voice or likeness. The open question, according to MBW, is whether the Act's clause on technology whose "primary purpose or function" is producing a person's voice can reach a general-purpose music generator. Both complaints allege that Suno's filter on artist names is easy to bypass, for example by prompting with style words only.

Suno says it does not "allow prompts for specific artists' names and songs" and screens uploads with third-party providers. On Nov 3 (window) Suno also asked a Boston judge to dismiss a proposed class action by Jason Isbell and three other musicians; the plaintiffs' response is due Dec 1 (upcoming).

Context, not window events: the Presley suit against Anthropic (October), Eight Mile Style v. Meta (filed May 2025, allowed to proceed to discovery) and earlier Suno suits by Round Hill, Reservoir, UMG and Sony. A suit by the American Federation of Musicians against Universal and Warner over their Suno and Udio licences is described, with a motion to dismiss and a union reply, only in an Instagram post by a music commentator; the court, filing date and status are not in the material, so it is listed as unverified. No docket event in the Disney-led studios v. MiniMax case appears in the material.

6.2 Licensing deals and training-data policy

The window's licensing news is one alliance rather than a music-label deal. Adobe and Disney announced a multi-year alliance making Adobe Disney's "Official AI Innovation Partner," which gives Disney teams access to Firefly, Firefly Foundry, Premiere and Photoshop, and was placed at Adobe MAX. The relevant point for copyright is the model design: according to Adobe's post, Firefly Foundry builds private models "customized using Disney intellectual property," and Imagineers are only "experimenting" with a sketch-to-image model. Neither company disclosed value, length in years or exclusivity.

Licensing background for Suno is context: Warner Music Group settled last year and licensed Suno, which also has deals with BMG and Believe; a label-backed model released in September drew a further UMG and Sony suit. Suno raised $400 million in June at a $5.4 billion valuation, per the Los Angeles Times.

6.3 Platform labelling, disclosure and provenance

Rule Effective Who it binds Source status
EU AI Act disclosure duty for AI-generated "deep fake" content (Article 50), penalties up to EUR 15m or 3% of turnover Aug 2, 2026 (in effect, context) Deployers, including advertisers Per Latent; Sideshift gives the same date
Meta Commercial Terms "AI Features" clause: advertisers "solely responsible" for what Meta's AI creates or alters Oct 30, 2026 (in effect, context) Advertisers Per Latent
Google "How this ad was made" panel in My Ad Center Jul 9, 2026 (in effect, context) Google advertisers Per Latent
India IT Rules amendment: visible AI labels, C2PA metadata, 24-hour takedown for impersonation Notified Jan 15, 2026 (context) Intermediaries Per AIFOD citing IMPRI

No platform policy change dated Nov 3 to Nov 16 appears in the material. The platforms' own policy pages are not in the material, so everything here is "according to" commentators. Per Latent, YouTube asks creators to disclose realistic altered content and AI music as the main focus, TikTok has a mandatory advertiser disclaimer, and Meta's mandatory ad disclosure applies only to political and social-issue ads.

The one primary provenance source is OpenAI's Help Center: images from ChatGPT or the API carry Content Credentials (C2PA) and SynthID watermarks, audio carries SynthID, and Sora is discontinued. The page is undated and does not mention Images 2.6 (see Chapter 3). OpenAI warns that a screenshot or file conversion can strip metadata. Per Watermark.ws, OpenAI announced SynthID for images on May 19, 2026 and Google made its visible watermark optional in August 2026 (secondary, pre-window).

6.4 What is still unresolved

No court ruling on training-data liability appears in the window. The Suno complaints push the argument toward voice and likeness law, where an output need not copy a song to be challenged. Whether label licences reach session musicians is untested in the material. On platforms the gap is enforcement: provenance metadata is fragile, so a platform may not label what it cannot read. This is description, not legal advice.

7. Commercialization and the creator ecosystem

Chapter 6 showed rights holders pushing back on one music generator; the commercial side of the window shows money and distribution flowing toward companies that sit between models and creators.

7.1 Funding and valuations

One in-window round, and it is Chinese and open-source: Sand.ai, a Beijing video-generation startup, raised $200 million in a Series A+ co-led by CoStone Capital and Wangsu Science & Technology, with a valuation reported above $1 billion.

Company Round Amount Valuation Lead / participants Date Status
Sand.ai (MAGI models, VidMuse app) Series A+ $200 million Above $1 billion (reported) CoStone Capital and Wangsu Science & Technology (co-leads); Times Capital, Pudong Venture Capital, Shanghai Guofang and others Nov 5 per toflow.ai; reports dated about Nov 7 (window) Reported; round and investors per an investor post
fal (inference platform for image, video, audio models) Round led by Bessemer $1.3 billion or $1.65 billion (sources disagree) $16 billion Bessemer; Nvidia, Sound Ventures, T. Rowe Price October (context) Reported, social posts only
Higgsfield Series B extension Undisclosed $5.4 billion (as of Aug 2026) Not stated in this source October (context) Tracker estimate of announced terms
Runway Series E $315 million $5.3 billion Not stated Feb 2026 (context) Tracker

Three cautions apply. The Sand.ai figures come from an investor's social post and a sales-intelligence page, not the company or a filing; Sand.ai itself is quoted saying its VidMuse app passed $10 million in annual recurring revenue within two months of launch, which is a company claim. Wangsu also disclosed a planned 300 million yuan investment tied to inference and deployment cooperation; we do not convert currencies. Where two sources disagree on fal's October round, we follow neither: $1.3 billion and $1.65 billion both appear and the material has no company statement to settle it. The New Market Pitch tracker, updated Nov 14, ranks Higgsfield first at $5.4 billion, Runway at $5.3 billion and Decart, Luma AI and Synthesia at $4.0 billion each; it separates 24 valuations backed by statements or round terms from 52 estimates, which is the right habit to copy.

7.2 Distribution and aggregators

The window's distribution moves share one pattern: the model maker sells through someone else's surface.

Who sells whose models: Adobe Premiere Generative Media, Runway Dev, ElevenMusic to SoundCloud and licensed label music in Premiere mobile.
Who sells whose models: Adobe Premiere Generative Media, Runway Dev, ElevenMusic to SoundCloud and licensed label music in Premiere mobile.
  • Adobe as an aggregator. At Adobe MAX (Nov 10) Premiere's Generative Media tool lets editors choose Adobe Firefly or partner models from Google, Kling, Runway and Luma, and Google's latest imaging and video models are due to come to Firefly, Photoshop, Premiere and Express (product detail in Chapter 3). The Adobe-Disney alliance is the studio-side counterpart (Chapter 6).
  • Developer platforms reselling video models. Runway Dev sells the official Kling 4.0 API self-serve, alongside its own models and a Model Router that picks a model by cost, quality or latency; LiteLLM lists Kling 4 on Runway at $0.1256 per second as of Nov 13 (Chapter 5).
  • Google's customer list as distribution. Google's Cloud post names Artlist, FLORA, invideo, Krea, Picsart, Photoroom and Runway as builders on Gemini Omni 1.2 Flash; these are vendor-quoted testimonials.
  • Music distribution. ElevenMusic output can be published directly to SoundCloud, and Adobe's free Premiere mobile app adds licensed UMG and WMG music (Chapter 3).

7.3 Creator economics

Consumer plan prices in the material are third-party lists, so they are directional: a comparison captured in Oct 2026 (before the window) puts Higgsfield Plus at $39 per month billed annually with a credit price beside each listed video model, and Hailuo AI's Standard plan at $8.40 per month billed yearly. Adobe also announced $1 million in direct grants to 10 filmmakers through its Film & TV Fund on Nov 9, on top of nearly $10 million committed in grants and donated products for 2026. A Nov 13 social post reports that Disney plans to launch a preschool series made with agentic AI tools in December, with artists reviewing every frame; the source is a social post and is not verified.

7.4 What the money signals

Pricing power is moving to whoever owns the creator's workflow. The cheapest models are now one click away inside tools that already have the audience (Premiere, Firefly, Runway Dev), and valuations at the top of the tracker cluster at $4.0 to $5.4 billion for workflow and product companies, not for model labs alone. The one fresh round, Sand.ai's, pairs an open-source model with a consumer app. For creators the implication is portability: a model can be repriced, rerouted or left to decay (see the Grok Imagine report in Chapter 2), so keep prompts, references and edits outside any single model. Nothing here is investment advice, and tracker valuations marked "estimate" are not company statements.

8. Trend assessment and what to watch next

Chapters 2 to 7 covered models, access, price, rules and money; this chapter reads them together and lists what to watch in the next 30 days.

Capability news in the window was concentrated in two releases (Gemini Omni 1.2 Flash and OpenAI Images 2.6); the larger story is that models are reaching creators through other companies' surfaces, at prices that differ by seller, under licences and lawsuits that decide where they may be used.

8.1 Trends and how strong the evidence is

Trend Evidence in this window Confidence
Longer, cheaper production clips Omni 1.2 Flash 15 s at $0.07 per second; Kling 4.0 Flash 3 to 20 s; AA puts the top five within 23 Elo while prices span $4.80 to $34.12 per minute Strong
Clips to interactive world models Atlas streamed at up to 15 fps as a research preview; no clip length, price or ranking to compare Weak
Aggregators alongside model makers Premiere offers Google, Kling, Runway and Luma models; Runway Dev sells Kling 4.0; Google's launch customers include Runway, Krea and Picsart Strong
Licensing and litigation in parallel Two new Suno suits on Nov 6; UMG and WMG music licensed into Premiere mobile; ElevenMusic to SoundCloud Mixed
Open weights with strings attached H3-Image weights excluded in US, EU, UK, South Korea; LTX-2.5 free below $10M revenue; Wan 3.0 closed Mixed
Single-model dependence is a risk Grok Imagine team cut, product still live; Veo 3.1 previews gone from the Gemini API Mixed

The strongest signal is price dispersion: the same Kling 4.0 Flash costs $0.094 or $0.1174 per second depending on the seller, and the top-ranked model is not the cheapest. The weakest is the world-model story, which rests on one vendor demo.

8.2 What it means by reader

Reader What changed Implication Evidence
Solo creator Plans differ by seller: Runway plans work out at $0.096 to $0.230 per Gen-4.5 second; Omni 1.2 Flash is $0.07 per second at 720p Compare cost per second, not plan price; keep prompts and references outside any one model —
Studio or agency Adobe puts several makers' models in one timeline; Adobe-Disney alliance uses private models on Disney IP Pick a surface that lets you switch models; ask vendors for licence and provenance terms in writing —
Developer integrating APIs Omni 1.1 Flash reroutes to 1.2 on Dec 3; Vertex Veo 3.1 retirement floor is Nov 17 Pin model IDs, test the new price, and read the lifecycle page before cutover —
Rights holder Two Suno complaints (Nov 6) test the ELVIS Act and copyright metadata claims; response due Dec 1 in the class action Dated procedural steps exist, but no ruling on training is scheduled in the material —

These are readings of the evidence, not legal or investment advice.

8.3 Key dates

Dates up to Nov 17, 2026 are in effect or the matching status word; later ones are upcoming.

Date Event State on Nov 17, 2026 Source
Oct 22, 2026 Gemini API Veo 3.1 preview models shut down (context) In effect Google AI for Developers
Nov 3, 2026 Suno motion to dismiss the Isbell class action (Boston) Filed MBW
Nov 5, 2026 Gemini Omni 1.2 Flash GA In effect Google
Nov 6, 2026 Presley rights holders and Eight Mile Style file against Suno Pending MBW
Nov 10, 2026 Adobe MAX announcements Announced Adobe
Nov 17, 2026 Vertex veo-3.1-generate-001 retirement date "or later"; MiniMax H3-Image half-price 4K promotion ends (per CellCog) Upcoming or in effect on the run date; check the pages Google Cloud docs; CellCog
Nov 30, 2026 Runway Dev 20% Kling 4.0 promotion ends Upcoming Runway Dev
Dec 1, 2026 Plaintiffs' response due to Suno's motion to dismiss (Isbell class action) Upcoming MBW
Dec 3, 2026 gemini-omni-1.1-flash requests route to Omni 1.2 Flash Upcoming Gemini API release notes
Early Dec 2026 (estimate) Midjourney V9, "within about 30 days of Nov 4" per secondary summaries Upcoming, not confirmed kie.ai

9. References

This document draws on the following 67 sources, listed in the order they are first cited.

  1. Gemini Omni 1.2 Flash: 15-second video and lower pricing — blog.google
  2. Gemini Omni 1.2 Flash now GA: cost-effective video | Google Cloud Blog — cloud.google.com
  3. Kling AI on X: "Kling 4.0 Flash Is Now Fully Available! ⚡ Kling 4.0 Flash is now available to everyone on the Kling AI website and via API. You can also access it through MCP, CLI, and AI plugins, bringing greater flexibility to your video creation workflows. With Kling plugins for professional c… / X — x.com
  4. Frameboy: Atlas in Real-time | World Labs — worldlabs.ai
  5. Introducing Images 2.6 — openai.com
  6. What Is Midjourney v9? Specs, Changes, Roadmap — kie.ai
  7. Google AI Studio — ai.google.dev
  8. Model versions and lifecycle  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentation — docs.cloud.google.com
  9. Veo 3.1  |  Gemini Enterprise Agent Platform  |  Google Cloud Documentation — docs.cloud.google.com
  10. MiniMax H3-Image: 16M-Pixel Images, Prices, Open Weights | CellCog — cellcog.ai
  11. AA-Video-T2V v2.1 Leaderboard - Top AI Video Models | Artificial Analysis — artificialanalysis.ai
  12. AA-Image-T2I v2.1 Leaderboard - Top AI Image Models | Artificial Analysis — artificialanalysis.ai
  13. AI Video, Image and Audio API Pricing | Runway Dev — dev.runwayml.com
  14. Kling 4.0 Flash API: Pricing and Specs | Venice — docs.venice.ai
  15. Elvis Presley rightsholders sue Suno, alleging its AI imitates the King’s voice. Eight Mile Style has filed a separate suit over 269 Eminem songs. - Music Business Worldwide — musicbusinessworldwide.com
  16. Elvis rights holders and Eminem’s publisher sue Suno, testing the ELVIS Act — spokesman.com
  17. Sand.ai raised $200 million in a Series A+ round to build video generation models and tools for creators. CoStone Capital and Wangsu Science & Technology co-led the round. Sand.ai takes two routes… | Grishin Robotics — linkedin.com
  18. What Funding Has Sand.ai Raised? Full Round History | toflow.ai | toflow.ai — toflow.ai
  19. Explore new ways to create across Creative Cloud and Adobe Firefly while advancing your craft with more precision, choice and control — blog.adobe.com
  20. ElevenLabs now lets users publish AI-generated music directly on SoundCloud — tech.yahoo.com
  21. Premiere mobile adds music from UMG and WMG, helping creators find the perfect track for every story — blog.adobe.com
  22. Adobe and Google Deepen Partnership to Bring Leading AI Models and Integrated Workflows to Creatives and Enterprises — news.adobe.com
  23. AI video news November 2026: Kling 4.0, GPT Image 2.6 | AI-0 — ai-0.net
  24. ElevenLabs and NVIDIA expand voice agents for enterprise — elevenlabs.io
  25. Grok Imagine Loses Most of Its Video Team: Keep Your Film Portable — screenweaver.ai
  26. Kling 4.0 API: access, pricing and code | Runway Dev — dev.runway.com
  27. GitHub - useapi/kling-api: Kling API examples by useapi.net: Kling 4.0, 3.0 and Omni video (text-to-video, image-to-video with keyframes, multi-shot, motion control, native sound), lip-sync avatars, KOLORS images and a hosted MCP server (45 tools), on your Kling AI subscription, with no Kling developer account. · GitHub — github.com
  28. MiniMax H3 (Hailuo 3) AI Video: 2K Clips with Native Sound — ocmaker.ai
  29. Atlas | World Labs — worldlabs.ai
  30. MCP server · Atlas — docs.atlas.worldlabs.ai
  31. ChatGPT Images 2.6 - OpenAI Deployment Safety Hub — deploymentsafety.openai.com
  32. 🌹🦋 Timeless Visions AI Art Challenge Alerts 🌹🦋 | 🚀 MIDJOURNEY V9 IS GETTING CLOSER | Facebook — facebook.com
  33. Images, upgraded—introducing ChatGPT Images 2.6 - Announcements - OpenAI Developer Community — community.openai.com
  34. ElevenLabs’ new app combines meeting note-taking and dictation for teams — techcrunch.com
  35. ElevenLabs SoundCloud Deal: What We Know (Nov 2026) — mubert.com
  36. Google AI Studio — ai.google.dev
  37. Video generation in the Gemini API  |  Google AI for Developers — ai.google.dev
  38. Changelog | OpenAI API — platform.openai.com
  39. Deprecations | OpenAI API — platform.openai.com
  40. Best Local Video Generation Models 2026: VRAM and Licence — vdf.ai
  41. Best AI Video Generation Models in 2026: Elo, Price per Second, Open Weights - MarkTechPost — marktechpost.com
  42. AI 3D Model Generators for Game Development (2026) | Wavect — wavect.io
  43. Ranking AA-Video-T2V v2.1 - Principais modelos de vídeo com IA | Artificial Analysis — artificialanalysis.ai
  44. Video Arena Is Live on Web - Arena.ai — arena.ai
  45. How to Generate Videos with Video Arena | Arena Help Center — help.arena.ai
  46. AA-Video-Editing v1.1 Leaderboard - Top AI Video Models | Artificial Analysis — artificialanalysis.ai
  47. AA-Image-Editing v2.1 Leaderboard - Top AI Image Models | Artificial Analysis — artificialanalysis.ai
  48. Kling 4 API Pricing (Nov 2026): $0.1256 per Second | LiteLLM — models.litellm.ai
  49. AI Image and Video Pricing from $12/month | Runway AI — runwayml.com
  50. Gemini Omni Flash - Gemini API | Google AI Studio — ai.google.dev
  51. Creating with Gen-4.5 – Runway — help.runwayml.com
  52. Elvis Presley, Eminem's rights holders sue AI company Suno - Los Angeles Times — latimes.com
  53. Elvis Presley rights holders sue AI music generator over licensing, copyright violations | Courthouse News Service — courthousenews.com
  54. Eminem's Publisher Sues AI Music Generator Suno For $40M — vibe.com
  55. Instagram — instagram.com
  56. Adobe gains marketing tie-ins with 5 Disney units in multi-year AI alliance — ppc.land
  57. Is AI-generated advertising allowed? The 2026 rules | Latent — thesillyhumans.com
  58. AI Labels Explained: What They Mean for UGC and Ad Performance - Sideshift Blog — sideshift.app
  59. India Mandates 24-Hour Deepfake Takedowns and C2PA Labels | AIFOD — af.net
  60. Provenance signals in content processed by OpenAI tools | OpenAI Help Center — help.openai.com
  61. Do AI Images Have Invisible Watermarks? | Watermark.ws — watermark.ws
  62. [𝐟𝐚𝐥 𝐑𝐚𝐢𝐬𝐞𝐬 $𝟏.𝟑 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐚𝐭 𝐚 $𝟏𝟔 𝐁𝐢𝐥𝐥𝐢𝐨𝐧 𝐕𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧

fal is a startup that hosts and runs cutting-edge AI video, image, and audio models. Designed to empower developers and creators with high-performance infrastructure, the company provides scalable APIs and fast tools that eliminate deployment bottlenecks, making it easy to integrate state-of-the-art generative AI into real-time applications.

fal has raised $1.3 billion in its latest funding round, bringing the company’s valuation to $16 billion. The round was led by Bessemer Venture Partners, with participation from Nvidia, Sound Ventures, and T. Rowe Price.

fal, headquartered in San Francisco, California, United States, was founded in 2021 by Burkay Gur and Gorkem Yurtseven. In 2025, the company joined the unicorn club. “This year, we at fal pushed the limits of what’s possible, of what’s imaginable, in generative media. We post-trained already outstanding models to be even better and faster, in order to test what happens when quality converges with speed. We also cemented generative media’s impact in every major industry, expanding the very definition of what a creative layer can enable,” wrote the company in its post. | Silicon Valley Investclub](https://www.facebook.com/siliconvalleyinvestclub/videos/%F0%9D%90%9F%F0%9D%90%9A%F0%9D%90%A5-%F0%9D%90%91%F0%9D%90%9A%F0%9D%90%A2%F0%9D%90%AC%F0%9D%90%9E%F0%9D%90%AC-%F0%9D%9F%8F%F0%9D%9F%91-%F0%9D%90%81%F0%9D%90%A2%F0%9D%90%A5%F0%9D%90%A5%F0%9D%90%A2%F0%9D%90%A8%F0%9D%90%A7-%F0%9D%90%9A%F0%9D%90%AD-%F0%9D%90%9A-%F0%9D%9F%8F%F0%9D%9F%94-%F0%9D%90%81%F0%9D%90%A2%F0%9D%90%A5%F0%9D%90%A5%F0%9D%90%A2%F0%9D%90%A8%F0%9D%90%A7-%F0%9D%90%95%F0%9D%90%9A%F0%9D%90%A5%F0%9D%90%AE%F0%9D%90%9A%F0%9D%90%AD%F0%9D%90%A2%F0%9D%90%A8%F0%9D%90%A7fal-is-a-startup-that-hosts-and-r/1074828878850229) — facebook.com 63. Instagram — instagram.com 64. What are the most valuable AI video generation startups? – New Market Pitch — newmarketpitch.com 65. 10 Google Flow Alternatives in 2026: Veo, Credits & Cost — saascrmreview.com 66. Adobe backs the next generation of filmmakers with $1 million in direct funding and new industry Advisory Board — blog.adobe.com 67. Ventureburn — facebook.com

Interactive web page · scroll inside the frame Open full page ↗

Generated with Divrite — this work was produced by AI from the prompt above and may contain mistakes. Check the cited sources before relying on it.

Related features

Ask your own question

Divrite researches it in depth, checks the key figures against their sources, and turns it into a deck, report or web page.