OpenAI Images 2.6: GIF Creation and Two New API Models
OpenAI's Images 2.6 adds GIF generation to ChatGPT and Codex and brings Flare and Sunburst models to the API at the same token rates as 2.5.
OpenAI's Images 2.6 adds GIF generation to ChatGPT and Codex and brings Flare and Sunburst models to the API at the same token rates as 2.5.
Introduction
OpenAI introduced Images 2.6 on November 4, 2026, an update to the image model behind ChatGPT, Codex and the API. The headline addition is GIF generation: ChatGPT can now produce an animated loop from a text prompt or an uploaded photo, something The Verge notes it could not do natively in a single step before. OpenAI also says Images 2.6 "generates images 50% faster than before," and it ships two models to its API, GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst, together with new controls for image markup, SVG export and a "complexity" setting.
The update follows Images 2.5, which launched in September. OpenAI says people now create 4 billion images per week across ChatGPT, Codex and the API, which works out to more than half a billion per day. The Verge notes that the company reported 3 billion per week in September.
Feature Overview
GIF generation
OpenAI describes several uses for the new capability: animating a mascot or character, turning a product shot into a loop, and building animated storyboards. A simple pen sketch can also become an animated character. According to The Verge, a GIF can start from a prompt or from a photo uploaded to the chatbot, and the result can be adjusted by animation speed, edited frame by frame and shared. The example OpenAI supplied was a prompt asking for a GIF of a cat dancing, with controls to change playback speed and step through frames. The sample GIFs in OpenAI's post are credited to OpenArt.
Speed and the Flare model
OpenAI's wording is that Images 2.6 generates images 50% faster than before. The announcement does not name a baseline model beyond "before," so the figure is best read as a directional claim. A more concrete number applies to the API: GPT-Image-2.6 Flare, with quality set to low, can generate an image in as little as six seconds.
Two API models
The API gets two models. Flare is the speed-oriented option. Sunburst is described as the higher-quality option, with greater consistency across edits, meaning characters, products and brand details stay consistent as an image is revised. The same Flare and Sunburst naming was used for Images 2.5, so this release continues an existing split rather than introducing a new one.
New API controls
Three additions target developers who build editing and design tools.
| Control | What it does |
|---|---|
| Image markup | Already available in ChatGPT, now in the API, so a request can pinpoint where a change should happen |
| SVG export | Lets the API return a vector file |
| Complexity parameter | Sets how much planning the model does before generating |
OpenAI recommends low complexity for most edits and medium or high for scenes with many elements or constraints. Quality is a separate setting that governs detail and fidelity, while complexity guides how much goes into the scene.
Availability and pricing
Images 2.6 is available globally in ChatGPT and Codex across all subscription tiers, on desktop, mobile and web. In the API, OpenAI's pricing page lists the two new models at the same standard rates as the Images 2.5 models, per 1M tokens.
| Model | Image input | Image output | Image cached input | Text input | Text cached input |
|---|---|---|---|---|---|
| gpt-image-2.6-sunburst | $8.00 | $30.00 | $2.00 | $5.00 | $1.25 |
| gpt-image-2.6-flare | $8.00 | $30.00 | $2.00 | $5.00 | $1.25 |
| gpt-image-2.5-sunburst | $8.00 | $30.00 | $2.00 | $5.00 | $1.25 |
| gpt-image-2.5-flare | $8.00 | $30.00 | $2.00 | $5.00 | $1.25 |
The batch price table lists only 2.5 and earlier models.
Usability Analysis
For ChatGPT users, the practical change is that animation sits inside the same conversation as image creation. A prompt or photo goes in, a GIF comes out, and speed and individual frames can be edited without leaving the chat. The use cases OpenAI highlights, a looping product shot, a mascot animation or a storyboard, are the kind of short assets marketers and designers produce routinely.
For developers, the interesting piece is the pairing of two settings. Quality and complexity are independent, so a team can keep complexity low for routine edits and raise it for crowded scenes, while choosing Flare for quick drafts and Sunburst when consistency across revisions matters. Markup in the API means an editing UI can send a precise region rather than describing the location in words. SVG export extends the output beyond raster images.
OpenAI also cites an early-access example: Town, a startup building an AI work assistant, used Images 2.6 in the API to explore animated characters.
Pros and Cons
Pros
- GIF creation, speed control and frame editing in one place
- Flare and Sunburst give a clear speed versus consistency choice
- Quality and complexity are separate API levers
- Per-token prices match Images 2.5
- Wide availability across ChatGPT and Codex tiers and platforms
Cons
- The six-second Flare figure is tied to the low quality setting
- The 50% speed claim names no baseline
- Token rates did not fall with the new release
- The batch price table does not yet include 2.6 models
Outlook
Image generation is increasingly measured by workflow fit rather than single-image quality, and Images 2.6 reflects that: markup, SVG export and the complexity setting are all about controlling edits. The jump from 3 billion to 4 billion weekly images between September and November, as reported by OpenAI and The Verge, shows how quickly volume is growing. Items worth watching are whether the batch price table adds 2.6 rows and what developers report about Flare and Sunburst in production.
Conclusion
Images 2.6 is a practical release. GIF generation fills a gap in ChatGPT, and the API additions give developers finer control over speed, consistency and planning. The headline speed figures come with qualifiers, and pricing is unchanged from Images 2.5. It suits creators who want short animated assets from a prompt and developers building image editing tools, and it is a smaller step for anyone who only needs still images. Rating: 3.5 out of 5.
Editor's Verdict
OpenAI Images 2.6: GIF Creation and Two New API Models brings real, demonstrable value, though with caveats that deserve weighing.
The strongest case for paying attention: the GIF workflow covers prompt, photo upload, speed adjustment and frame editing in one place, so a loop can be created and refined without leaving the chat. Reinforcing that, a split between Flare and Sunburst lets developers pick speed or consistency for each workload — practical value rather than just headline appeal. The broader signal worth registering is straightforward: the GIF feature puts animation in the same chat flow as image creation, with speed and frame editing handled inside ChatGPT according to The Verge. On the other side of the ledger, one constraint is real rather than a marketing footnote: the six-second Flare figure applies to the low quality setting, so the headline speed is tied to the lowest-detail option. It should factor into any serious decision. Layered on top of that, the 50% speed claim names no baseline beyond the phrase faster than before, which makes it hard to compare with earlier releases — which narrows the set of teams for whom this is an obvious yes.
For creators who need short animated assets and developers building image-editing tools on the OpenAI API, a measured trial makes sense, ideally comparing Flare and Sunburst at the quality and complexity settings a real workload needs.
Pros
- The GIF workflow covers prompt, photo upload, speed adjustment and frame editing in one place, so a loop can be created and refined without leaving the chat
- A split between Flare and Sunburst lets developers pick speed or consistency for each workload
- Separate quality and complexity settings give API users two distinct levers for fidelity and scene planning
- Token rates match Images 2.5, so the standard per-token prices stay the same as the previous generation
- Availability spans all subscription tiers in ChatGPT and Codex on desktop, mobile and web
Cons
- The six-second Flare figure applies to the low quality setting, so the headline speed is tied to the lowest-detail option
- The 50% speed claim names no baseline beyond the phrase faster than before, which makes it hard to compare with earlier releases
- The batch price table lists only 2.5 and earlier models, so batch pricing for the 2.6 models is not shown there
- Token rates did not fall with the release, so the speed gain does not lower the published cost
References
Comments0
Key Features
1. GIF generation in ChatGPT and Codex from a prompt or an uploaded photo, with adjustable animation speed and frame-by-frame editing. 2. Images generated 50% faster than before, per OpenAI; GPT-Image-2.6 Flare can produce an image in as little as six seconds at low quality. 3. Two API models: Flare for speed and Sunburst for higher quality and consistency across edits. 4. API additions: image markup, SVG export and a complexity parameter that sets how much planning the model does. 5. Standard API pricing identical to the Images 2.5 models: $8.00 image input and $30.00 image output per 1M tokens.
Key Insights
- The GIF feature puts animation in the same chat flow as image creation, with speed and frame editing handled inside ChatGPT according to The Verge
- Two API models split the work, with Flare aimed at speed and Sunburst at quality and consistency across edits
- Quality and complexity are independent settings, so detail and fidelity can be tuned separately from how much planning the model does
- Image markup arriving in the API lets editing tools send a precise region to change instead of describing it in words
- SVG export extends API output beyond raster images
- Standard token pricing matches the Images 2.5 models, so the 2.6 gains do not come with a published price change
- OpenAI reports 4 billion images per week across ChatGPT, Codex and the API, up from the 3 billion per week reported in September
- Early access use by Town, an AI work assistant startup, centered on animated characters
Was this review helpful?
Share
Related AI Reviews
OpenAI Hardens GPT-6.2 Reasoning Against Distillation
OpenAI says updated safeguards cut red-team extraction of GPT-6.2 Astra's protected reasoning to 0.2%, from 54.0% on GPT-6 Astra.
ChatGPT Wallet: Agent Payments With Per-Purchase Approval
OpenAI's ChatGPT Wallet lets dots agents pay with payment tokens after a per-purchase security check, starting with US Pro subscribers.
GPT-6 Cyber: 48.5% on ExploitGym, Built to Stay in Scope
OpenAI's GPT-6 Cyber scores 48.5% on ExploitGym and avoided out-of-scope targets in 99.93% of honeypot tests. Access starts with select customers.
OpenAI Trains Frontier Models Offline After Red-Teaming
OpenAI says frontier training now runs with no live internet access, after agent red-teaming found microVM escapes and covert channels.
