HaiAI123

Curated Global AI Tools Directory

Industry update
↓ 24%

MiniMax rolls out H3 Image for e-commerce and ad creative production

11/10/2026 — 11/12, 21:20·8 sources (all)·17 reports (all)·In the 2026-11-11 briefing

Key takeaways

  • MiniMax has rolled out H3 Image, an image model that generates 4K images in about 10 seconds, accepts up to 16 reference images per generati…

Background & analysis

Late on November 10, 2026, MiniMax announced the release of H3-Image, an image generation model derived from H3, and opened its weights. The model supports text-to-image generation, image reference, and image editing. MiniMax says it can produce a 4096×4096 pixel, roughly 16MP (4K) image within 10 seconds, about 5 times faster than mainstream options, and says the post-training pipeline was strengthened in areas including data, reinforcement learning, step compression, PE, and deployment optimization. H3-Image was made available through MiniMax's API, MiniMax Code, and Hailuo AI, with a launch-week 50% discount on 16MP and higher resolutions reported by coverage in China.

On November 11, H3-Image appeared on Hugging Face. MiniMax's official account then disclosed a joint launch with fal and noted that Runware offered Day 0 support, describing the model as accepting 16 reference inputs and running 2–3x faster, aimed at ad creative, e-commerce, and large-scale content production. A developer also launched a web-based generation tool built on H3-Image, letting users choose aspect ratio, 1K/2K/4K resolution, quality, and the number of images. Some outlets described H3-Image as an open SOTA 4K image model, and one report put the per-image cost at roughly two yuan, without sharing benchmark results or model parameters. A DEV Community post said the open-weights license had sparked discussion, and accounts of the terms did not yet agree.

That evening, MiniMax showcased H3-Image outputs including documentary sports photography and surrealist dreamcore images, shared the accompanying prompts, and invited users to submit their own work. It also explained that H3-Image can turn complex information and long case narratives into diagrams, posters, and evidence boards with timelines and annotations, supports up to 16 reference images in a single generation for asset layout, spatial composition, and product packaging, and can follow precise editing instructions. On November 12, MiniMax positioned H3 Image for e-commerce and ad creative at scale, emphasizing consistent multi-scene generation from a single sketch while keeping product features stable across frames. Public information currently stops at the release, weight opening, and channel rollout, with the licensing dispute unresolved.

AI-generated from 17 reports · updated 22 hours ago

Latest turnMiniMax has open-sourced an image generation model it positions as SOTA, featuring native 4K output and a cost of roughly CNY 0.2 per image, according to QbitAI. The coverage highlights a "CAD teardown" angle, though no benchmark numbers or model specs were included.

updream
24-hour heatHeat index 78 · peak 370 · 23h ago
24 hours agonow

Reports on this story headlines open the original

Yesterday
  1. MiniMax has open-sourced an image generation model it positions as SOTA, featuring native 4K output and a cost of roughly CNY 0.2 per image, according to QbitAI. The coverage highlights a "CAD teardown" angle, though no benchmark numbers or model specs were included.

    量子位AI score 63
  2. MiniMax has introduced H3 Image, designed to generate consistent multi-scene visuals from a single product sketch. The model targets e-commerce and ad creatives at scale while maintaining structural consistency across frames.

    X:MiniMax (@MiniMax_AI)First-partyAI score 60
Nov 11
  1. MiniMax officially announced that its image generation tool now supports up to 16 reference images in a single generation. This capability is useful for asset arrangement, spatial composition, and product packaging design.

    X:MiniMax (@MiniMax_AI)First-partyAI score 61
  2. MiniMax officially released H3-Image, a model capable of converting complex information and lengthy case narratives into diagrams, posters, and evidence boards with timelines and annotations.

    X:MiniMax (@MiniMax_AI)First-partyAI score 66
  3. MiniMax officially shared surrealist photography and corresponding prompts generated by creators using H3-Image, featuring a blend of Dreamcore aesthetics and minimalist conceptual photography.

    X:MiniMax (@MiniMax_AI)First-partyAI score 42
  4. MiniMax officially showcased cinematic dynamic photography and prompts created by creators using H3-Image, blending slow-shutter panning with classic 35mm color film aesthetics.

    X:MiniMax (@MiniMax_AI)First-partyAI score 42
  5. MiniMax officially showcased creation results using the H3-Image model and invited users to share their own generated images and prompts in the replies.

    X:MiniMax (@MiniMax_AI)First-partyAI score 50
  6. According to TestingCatalog, Brevis has launched Arca, a free app allowing users to sell local AI chat and coding sessions after automatically removing identifiable information like names.

    X:TestingCatalog (@testingcatalog)AI score 68
  7. According to reports, MiniMax has launched H3-Image, an open-weights image generation model built on its H3 video architecture. It generates native 4K images in about 10 seconds and supports 16 reference images in a single edit, though its license has sparked discussion.

    DEV Community · AIAI score 69
  8. Runware announced Day 0 support for MiniMax H3 Image. The model supports 16 reference inputs and delivers 2–3x faster generation speeds, designed for ad creatives, e-commerce, and high-volume content workflows.

    X:MiniMax (@MiniMax_AI)First-partyAI score 57
  9. MiniMax has partnered with fal to launch H3 Image, capable of generating 4K resolution images in about 10 seconds. The model follows complex prompts, enables precise editing, and handles game and 3D assets to power agents and productivity tools.

    X:MiniMax (@MiniMax_AI)First-partyAI score 74
  10. According to reports, MiniMax has open-sourced its 4K image generation model, H3 Image. The model natively outputs 4096×4096 resolution images in under 10 seconds, reaching open-source SOTA performance.

    掘金AI score 72
  11. A developer on V2EX has launched an online image generation tool powered by MiniMax H3 Image. The tool supports text-to-image creation and reference image editing, allowing users to customize aspect ratios, resolutions, and output counts.

    V2EXAI score 66
  12. MiniMax H3-Image is now available on Hugging Face, MiniMax Code and the MiniMax API, according to TestingCatalog. It is an open-weight image model supporting 4K images, and the account notes that good open image models are scarce, making any new release a notable step.

    X:TestingCatalog (@testingcatalog)AI score 59
  13. MiniMax has released and open-sourced H3 Image, an image generation model that outputs up to 4096x4096 and, per the company, finishes a 4K image in roughly 10 seconds with up to 16 reference images. Weights are open, the model is available through MiniMax's API, MiniMax Code and Hailuo AI, and 16MP-plus generations are half price for the first week.

    华尔街见闻AI score 68
Nov 10
  1. MiniMax has launched H3-Image, an open-weight image model it says produces 16-million-pixel 4K images in 10 seconds, with precise prompt following, editing and support for up to 16 reference images. It is available now through MiniMax's API, MiniMax Code and Hailuo AI.

    X:MiniMax (@MiniMax_AI)First-partyAI score 55
  2. MiniMax has released H3-Image, an open-weights image generation model derived from H3 that supports text-to-image, image reference and image editing. The company says it can produce a 4096×4096 (16MP, 4K) image in under 10 seconds, roughly 5x faster than mainstream offerings, with post-training work aimed at better instruction following and multi-reference understanding.

    IT之家AI score 65

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →