OpenAI launched Images 2.6 on November 4, adding GIF creation, faster image generation, and more precise editing across ChatGPT and Codex as well as the developer API S¹. Within hours, a Hacker News thread on the launch had drawn 105 points and 83 comments S². The "faster" claim is OpenAI's own, with no independent benchmark published to confirm it.
My read: The two-model split is the detail worth watching. Flare handles speed. Sunburst handles editing fidelity. That separation tells me OpenAI is treating generation and editing as different engineering problems, not one model doing both. Until someone outside OpenAI measures generation latency against the previous version, "faster" is a marketing assertion, not a tested result.
Two models, two jobs
OpenAI's deployment safety system card, published the same day, names the underlying models as Sunburst and Flare P⁶. Flare is tuned for faster generation. Sunburst targets editing consistency and detail preservation, and responds better to sketches and image annotations P⁶. A community post on OpenAI's own forum confirms both are coming to the API P⁴.
The practical shift is in the editing workflow. A designer who today regenerates an image several times to fix one element, a misplaced hand or a wrong background colour, might get closer with a single annotated pass if Sunburst's sketch handling works as described.
Nobody outside OpenAI has tested that yet.
What the announcement leaves out
The launch post does not specify pricing or tier availability, and gives no regional rollout details S¹. It makes no claim about higher resolution than previous versions. It does not mention video output beyond GIFs.
Hacker News traction is real but shifting. One snapshot captured 105 points and 83 comments S², while another taken earlier the same evening recorded 92 points and 78 comments S³.
Point counts do not reveal whether developers are excited or sceptical.
Developers building on OpenAI's API should first check whether the new controls change existing rate limits or per-image pricing. OpenAI has not published those details. The most concrete technical document available so far is the system card on OpenAI's deployment safety hub, dated November 4, 2026 P⁶.
Sources: S1 — Introducing Images 2.6 · S2 — Images 2.6 · S3 — Images 2.6 · P4 — Images, upgraded—introducing ChatGPT Images 2.6 · P5 — huan-yin/Awesome-Image-Generation-Model · P6 — ChatGPT Images 2.6 System Card - OpenAI Deployment Safety Hub · P7 — Studio26 (Prints)
Written from 7 sourced items, 5 of them primary.
