AITOPAI热报
ModelsOfficialSelected73°

GPT-image-2.5 model released

官方一手歸藏(guizang.ai)@op7418
Post· 11 sources
View the post
Why we picked it

The model is now available at the same price as its predecessor with improved consistency for game asset creation.

GPT-image-2.5 has been released with faster image generation speed and improved reference consistency. The model maintains consistency during multiple edits and supports comment-based editing. Despite being version 2.5, the API pricing remains the same as GPT-image-2. The model shows reduced破碎感 (fragmentation) and涂抹感 (smudging) issues.

Image · 歸藏(guizang.ai)
Original · 歸藏(guizang.ai)

GPT-6 Astra + GPT-image-2.5 + JavaScript GPT-6 不只做 3D 游戏很厉害，做 2D 游戏也厉害 GPT-image-2.5 现在做游戏素材牛皮啊！ 角色的动作是用 GPT-image-2.5 生成的精灵图，然后做成的关键帧动作。 他自己做了：背景图、跳跃的平台、怪物、人物和怪物动作的精灵图、技能音效 Your browser does not support the video tag. 🔗 View on Twitter 歸藏(guizang.ai) @op7418 GPT-image-2.5 图像生成模型果然发布了。官方说明的优化点主要有这几个新特性： 更快的图像生成速度，更高的参考图一致性，多次编辑中的一致性保持，基于评论的编辑 目前已全量在 GPT 和 Codex 推出。但图像元数据里标注的依然是 2.0，所以无法通过元数据判断模型版本； 神奇的是，GPT-image-2.5 这两个版本的 API 价格跟 GPT-image-2 相比没有任何差异。 实际体验下来，破碎感和涂抹感的问题确实变好了一些，不过提升不算太大； 但一致性确实明显变强了，尤其是连续多次编辑的一致性。 🔗 View Quoted Tweet 💬 2 🔄 0 ❤️ 6 👀 1416 📊 3 ⚡

#GPT-image-2.5#OpenAI#image generation#game development#multimodal

Keep reading

Recent stories with the same tag or section
ModelsBuildersSelected00:43
HyperFlow H3 Flash: Slimmed MiniMax-H3 Runs on Consumer GPUs

Shrinking a video generation model to 9.1GB moves real-time 768P generation from data-centre hardware into the hands of hobbyists with a single consumer GPU. Developers and indie creators can now experiment with MiniMax-H3–class video models locally instead of renting cloud capacity. GitHub access means the community can benchmark it directly against heavier open-source video generators.

#HyperFlow H3 Flash#MiniMax#video generation#open-source models
Gorden Sun@Gorden_SunSource