Skip to content
← AI by field
Creators, studios, marketers and agencies

AI for content and media production, right now

Generating and editing images and video, writing, voice and music, and producing finished content.

No benchmark of its own yetUpdated Nov 12, 2026 from 9 sources
Open media in Spezia Free account. No card.
Where AI is in media now

AI now generates high-resolution images and editable 3D scenes from text, with MiniMax H3-Image and GPT Image 2.6 leading.

What models can do now

  • MiniMax H3-Image can directly generate 16-megapixel images in about 10 seconds without upscaling, and supports editing and up to 16 reference images.[7][8][9]
  • GPT Image 2.6 tops text-to-image benchmarks, with Flare and Sunburst variants taking first and second place on AA text-to-image and editing.[4]
  • Intangible's Canvas turns written briefs into editable 3D scenes with cameras, references, images, and video for AI video generation.[1]
  • Wan2.2-S2V transforms static images and audio into synchronized videos with dialogue, singing, and performance, supporting minute-level generation.[5]
  • fal offers LTX VFX video tools including alpha mattes, HDR conversion, reframing, and upscaling.[3]
  • A pDMD LoRA enables realtime two-step image generation and editing on MiniMax's 33B image model.[2]

Where they still fall short

  • MiniMax's claim of 10-second generation for 16-megapixel images has not been independently verified, and speed varies by hardware.[9]
  • Independent evaluations reveal endpoint and consistency gaps in GPT Image 2.6 despite its top benchmark scores.[4]

What is expected next

  • Gaxos expects ArtGen Muse to make high-quality AI video more accessible and economical, helping creators and businesses bring more ideas to life.[6]

How fast it is moving

The record on this field's benchmarks over the last year, from Spezia's model leaderboard.

Image and video models are rated by people's votes in LMArena's image and video arenas, shown below, rather than by test scores.

Images from a prompt

People's votes on LMArena (CC BY 4.0), 2026-11-09

  1. 1GPT-Image-2.5 Sunburst1,429
  2. 2GPT-Image-2.6 Sunburst1,406
  3. 3GPT-Image-2.5 Flare1,400
  4. 4GPT-Image-2.6 Flare1,383
  5. 5GPT-Image-21,382
  6. 6MAI-Image-2.61,329

Editing images

People's votes on LMArena (CC BY 4.0), 2026-11-09

  1. 1GPT-Image-2.5 Sunburst1,520
  2. 2GPT-Image-2.5 Flare1,484
  3. 3GPT-Image-2.6 Sunburst1,477
  4. 4MAI-Image-2.7-Preview1,468
  5. 5GPT-Image-21,461
  6. 6GPT-Image-2.6 Flare1,460

Video from a prompt

People's votes on LMArena (CC BY 4.0), 2026-11-09

  1. 1Gemini Omni 1.2 Flash1,540
  2. 2Gemini Omni 1.1 Flash1,513
  3. 3gemini-omni-flash1,509
  4. 4FLUX 3 Video1,491
  5. 5Grok Imagine Video 1.51,488
  6. 6Dreamina Seedance 2.01,483

Video from an image

People's votes on LMArena (CC BY 4.0), 2026-11-09

  1. 1Gemini Omni 1.2 Flash1,503
  2. 2Dreamina Seedance 2.51,494
  3. 3MiniMax H31,493
  4. 4Gemini Omni 1.1 Flash1,487
  5. 5Wan 3.01,482
  6. 6Dreamina Seedance 2.01,480

Milestones

What became possible, newest first, each marked by how firm it is. Open one for the words it rests on.

Showndemonstrated with published results, a paper or an independent checkClaimedsaid by the company or people who built it, not yet checked by othersReportedreported by others, not confirmed by the company or a published resultNot publicdone by a model that is not released to the public

  1. Nov 11, 2026ShownIntangible Canvas
    Intangible's Canvas turned written briefs into editable 3D scenes for AI video generation.
    “Intangible's Canvas turns written briefs into editable 3D scenes for AI video generation.” Intangible's Canvas turns written briefs into editable 3D scenes for AI video ge
  2. Nov 10, 2026ShownMiniMax-H3-Image
    MiniMax released H3-Image, an open-weights model that directly generates 16-megapixel images in 10 seconds.
    “MiniMax-H3-Image, an AI image generation tool capable of producing 16-megapixel images, has been released.” MiniMax-H3-Image, an AI image generation tool capable of producing 16-megapixel
  3. Nov 10, 2026ShownMiniMax-H3-Image
    A pDMD LoRA enabled realtime two-step image generation and editing on MiniMax's 33B image model.
    “A pDMD LoRA enables realtime two-step image generation and editing on MiniMax's 33B image model.” A pDMD LoRA enables realtime two-step image generation and editing on MiniMax's
  4. Nov 10, 2026ShownLTX VFX
    fal launched LTX VFX video tools including alpha mattes, HDR conversion, reframing and upscaling.
    “fal launches LTX VFX video tools including alpha mattes, HDR conversion, reframing and upscaling.” fal launches LTX VFX video tools including alpha mattes, HDR conversion, reframi
  5. Nov 10, 2026ShownWan2.2-S2V
    Wan2.2-S2V became natively supported in ComfyUI for audio-driven video generation.
    “Wan2.2-S2V, the advanced audio-driven video generation model, is now natively supported in ComfyUI!” Wan2.2-S2V in ComfyUI: Audio-Driven Video Generation
  6. Nov 6, 2026ShownGPT Image 2.6
    GPT Image 2.6 topped text-to-image benchmarks according to independent evaluations.
    “GPT Image 2.6 tops text-to-image benchmarks as new independent evals reveal endpoint and consistency gaps.” GPT Image 2.6 tops text-to-image benchmarks as new independent evals reveal endp
Every milestone, story and model Pick a model to see what it can do in media.

Sources

  1. [1]Intangible's Canvas turns written briefs into editable 3D scenes for AI video generation.Spezia news · 2026-11-11
  2. [2]A pDMD LoRA enables realtime two-step image generation and editing on MiniMax's 33B image model.Spezia news · 2026-11-10
  3. [3]fal launches LTX VFX video tools including alpha mattes, HDR conversion, reframing and upscaling.Spezia news · 2026-11-10
  4. [4]GPT Image 2.6 tops text-to-image benchmarks as new independent evals reveal endpoint and consistency gaps.Spezia news · 2026-11-06
  5. [5]Wan2.2-S2V in ComfyUI: Audio-Driven Video GenerationWeb · 2026-11-10
  6. [6]Gaxos Launches ArtGen Muse AI Video Model as Company ScalesWeb · 2026-11-09
  7. [7]MiniMax-H3-Image, an AI image generation tool capable of producing 16-megapixel images, has been released. It can run locally and supports image editing and referencing multiple images.Web · 2026-11-10
  8. [8]MiniMax releases H3-Image, its strongest open-source image generation model: generates 16MP images in 10 seconds, with open weights, but its license excludes Europe, the US, the UK, and South Korea. - King of Computer MediaWeb · 2026-11-10
  9. [9]MiniMax unveils open-weights model that generates 16-megapixel imagesWeb · 2026-11-10

Other fields