OpenAI launches ChatGPT Images 2.5: latency cut up to 50%, adds sketch reference and on-image annotation editing
量子位 · wechat · 2026-09-09
OpenAI has released ChatGPT Images 2.5, its next-gen image generation model. Generation latency drops by up to 50% versus Images 2.0, with new xhigh and max quality tiers (max resolution stays 3840px) that improve detail, texture, lighting, and reference-subject fidelity.
The headline upgrade is controllable editing: multi-round edits now better preserve previously confirmed content and quality; you can annotate directly on the image (circle an object and write what to change); and a new @Sketch feature turns hand-drawn rough sketches into visual references for full generations. Generation progress percentage is finally shown.
Two API variants shipped: GPT-Image-2.5 Flare (balanced quality/editing/speed, recommended default) and GPT-Image-2.5 Sunburst (for ad assets and high-fidelity product imagery), both priced at $5/M text input tokens, $8/M image input, $30/M image output. Community tests praise improved material rendering, but note sketch proportions still drift and limbs occasionally fail more often.
Related event: OpenAI launches ChatGPT Images 2.5 with Flare and Sunburst API models(43 posts)→
More from Multimodal
- Runway App Studio Demos: Character Swap and Production-Ready Character Sheets — runwayml · 2026-10-27
- Higgsfield Launches Playable: One-Prompt Playable Ads — CurieuxExplorer · 2026-10-27
- Guy turns his girlfriend's selfie into a playable AI game ad about popping pimples — CurieuxExplorer · 2026-10-27
- Fibo 2 edit called unstable: even a color change alters the edited object — jd3k · 2026-10-27
- AIArtsy AI generates 3D cartoon-style sword fighting animation clip — Jealous-Associate-41 · 2026-10-27
- Seeking AI workflows for high-quality anime comics and colored manga pages — Worldly_Courage395 · 2026-10-27