statementSaid on 2026-11-10
模型最多支持输入 16 张参考图,重点提升复杂指令理解、多图参考及画面细节表现。
A statement by post author, said on 2026-11-10. Statements are as reported by the cited publishers — not verified by DecisionGateway.
模型最多支持输入 16 张参考图,重点提升复杂指令理解、多图参考及画面细节表现。
Discussion
Related claims
该模型支持文生图、图像参考与图像编辑,可在约 10 秒内生成 4096×4096 分辨率、约 1600 万像素的 4K 图像。post author · statementFrequently used subjects such as characters, pets and mascots can be saved to a reference library and called directly in later prompts to reduce inconsistency of the same figure across images.post author · statementTwo API models launched simultaneously: GPT-Image-2.6 Flare focused on generation speed and Sunburst focused on image quality and consistency across multiple edits.post author · statement稀宇科技发布 MiniMax H3 Image 图像模型并开放权重。post author · statement模型已上线 Hugging Face 及开放平台 API,并接入 MiniMax Code、海螺等产品。post author · statementChatGPT and the API now generate more than 4 billion images per week, up from 3 billion when Images 2.5 launched in September.OpenAI · statement