Editorial illustration for OpenAI Launches GPT-Image-2.6 Flare and Sunburst, Adds 6-Second API Generation
OpenAI's GPT-Image-2.6 Splits Into Flare and Sunburst
OpenAI Launches GPT-Image-2.6 Flare and Sunburst, Adds 6-Second API Generation
OpenAI pushed out Images 2.6 on Tuesday across ChatGPT, Codex and the API, splitting its flagship image model into two versions aimed at different jobs. GPT-Image-2.6 Flare handles quick iteration, OpenAI says it can return a result in as little as 6 seconds when quality is set to low. Sunburst trades speed for precision, built for detailed compositions and edits that need to stay consistent across multiple passes.
The update isn't limited to raw speed. OpenAI added GIF output, SVG export, markup-guided editing and a new complexity parameter, features aimed squarely at developers building production tools rather than casual prompt-and-done users. The company claims the whole system runs 50% faster than the prior release, though it hasn't published benchmark scores or a per-image cost breakdown to back that up.
Both models stay closed-weight and OpenAI-hosted, billed by token, so there's no self-hosting option for teams that want to run this themselves. For a company facing growing pressure from Midjourney and Google's Imagen line, the bet here is that developers will pick OpenAI's API for the editing tools and output formats, not just for who generates the sharpest image.
OpenAI just shipped Images 2.6 across ChatGPT, Codex and the API. For developers, the core release is GPT-Image-2.6, shipped as two API models: Flare and Sunburst. OpenAI team states that Images 2.6 generates images 50% faster than before.
Why this matters
The Flare and Sunburst split tells us OpenAI is finally admitting that one image model can't serve both a chatbot doing quick mockups and a designer doing final compositions. A 6-second generation time at low quality is a real number developers can build against, not a vague speed claim, and that matters if you're running image generation inside a live product loop rather than a batch job. SVG export and GIF output also push GPT-Image-2.6 toward actual production design work, not just illustration demos.
The markup-guided edits and complexity parameter suggest OpenAI wants finer control knobs rather than one-shot prompting, which is the direction serious tooling teams have been asking for. Still, this is closed weights, token-billed, and locked to OpenAI's own surfaces, so anyone comparing cost per image against open alternatives needs to run their own numbers before switching workflows. Worth watching: whether Sunburst's "stronger consistency" claim holds up across multi-step edits, since that's the part most image APIs still fumble in practice.
Related Reading
Browse all OpenAI news →
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
Related Topic OpenAI researcher quits, citing distrust over ad‑driven engagement metrics
LLMs & Generative AI
Related Topic OpenAI launches GPT-Image 1.5 with precise editing for enterprise visuals
LLMs & Generative AI
From Archives OpenAI buys Pierre Computer Company to help Git keep up with AI coding agents
From Archives AI PACs spent millions to shape Congress amid Democratic wave
From Archives Brennan: OpenAI's 2028 Election Policy Assumes Mass AI Adoption Not Yet Seen
Open Source
From Archives ChatGPT Now Natively Generates Animated GIFs and Edits Individual Frames
LLMs & Generative AI
Common Questions Answered
What are the two versions of GPT-Image-2.6 and what are they designed for?
OpenAI split GPT-Image-2.6 into two API models: Flare and Sunburst. Flare is optimized for quick iteration and can return results in as little as 6 seconds when quality is set to low, while Sunburst trades speed for precision and is built for detailed compositions and edits that need to stay consistent across multiple passes.
How much faster does GPT-Image-2.6 generate images compared to the previous version?
According to OpenAI's official statement, Images 2.6 generates images 50% faster than the previous version. This significant speed improvement enables developers to build real-time image generation into live product workflows rather than relying solely on batch processing.
What new export and output formats were added in the GPT-Image-2.6 update?
OpenAI added GIF output and SVG export capabilities to GPT-Image-2.6. These new format options push the model toward actual production design work, making it more suitable for designers and developers working on final compositions.
Why did OpenAI create separate Flare and Sunburst models instead of one unified image model?
OpenAI recognized that a single image model cannot effectively serve both fast-iteration use cases like chatbot mockups and precision-focused work like designer final compositions. By splitting into Flare and Sunburst, OpenAI can optimize each model for its specific use case, allowing developers and designers to choose the version that best fits their workflow needs.
Further Reading
- Introducing Images 2.6 - OpenAI
- ChatGPT Images 2.6 System Card - OpenAI Deployment Safety Hub
- Image prompting - OpenAI Developers
- Image generation - OpenAI Platform
