whatmodel.sh / image / GPT Image 2.5 Sunburst
GPT Image 2.5 Sunburst
↗ share
+ contribute
GPT Image 2.5 Sunburst is a closed-weights image model from OpenAI, released September 2026. Licence: proprietary.
Score within the LMArena text-to-image comparison set only.
$ how it compares
The image models ranked either side of GPT Image 2.5 Sunburst, out of 197 with a score, then more from OpenAI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 1 | GPT Image 2.5 Sunburst | 92±2 | api $0.013/image | ||
| 2 | GPT Image 2.5 Flare | 88±2 | api $0.013/image | ||
| 3 | GPT Image 2 | 85±1 | api $0.053/image | ||
| 4 |  MAI-Image-2.6 | 81±2 | comparison only | ||
| 5 |  Nano Banana 2.1 | 81±2 | api $0.034/image | ||
| more from OpenAI | |||||
| 14 | ChatGPT Image Latest (chatgpt-image-latest) | 74±4 | api $0.034/image | ||
| 17 | GPT Image 1.5 | 71±1 | api $0.034/image | ||
| 42 | GPT Image 1 | 61±1 | api $0.042/image | ||
$ what it's good at
text to image▂▅▅▃▄▄█▃▁▂▁▃ #1 of 126 * · 48% of its score · 2 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LMArena text-to-image | 1424.6 ✎ | ▂▂▃▄▇█▄▂▄▂·▃ beats 75/75 | 45% |
| GenExam (relaxed overall, GPT-5 judge) | 95.3% ✎ | ▆▆·▂▂····▂▂█ beats 13/13 | 2.5% |
image to image / editing▁▁·▃▂▃█▃▃▄▁▃ #1 of 62 * · 17% of its score · 1 test
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena image edit | 1523.6 ✎ | ▂▁▁▂▆▇█▄▆▅▁▃ beats 53/53 | 17% |
other tests2 not in a task · 34% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| VVRBench-Challenge accuracy (720 harder verifiable tasks) | 21.4% ✎ | ▄·▄···█·▄▄·▄ beats 6/6 | 27% |
| Rapidata Benchmark.ai image benchmark (overall Elo) | 1238.3 ✎ | ▁▁▁▄▂█▂▁▁▂▁▃ beats 34/34 | 7.0% |
$ where to run it
Through an API
api · fal.ai · $0.013/image · fal.ai ↗ · price sourceUSD per generated image, 1024×1024 medium quality: fal's per-size/per-quality table for ChatGPT Images 2.5 on this endpoint lists 1024×1024 at low $0.00588, medium $0.01317, high $0.05268, xhigh $0.09366, max $0.21072 (fal default quality is high). Product: fal.ai proxy of the OpenAI image model. API model: openai/gpt-image-2.5/sunburst/text-to-image (OpenAI gpt-image-2.5-sunburst). Settings: 1024×1024, quality=medium, num_images=1. Price basis: fal's own published per-image figure (fal bills tokens underneath at $5/1M text input and $8/$30 per 1M image input/output tokens; fal states longer or more complex prompts cost more, so input text tokens are additional). Not an OpenAI direct tariff (OpenAI publishes token prices only). Currency: USD. fal.ai official model page, observed 2026-09-25.
$ how the score was worked out · LMArena leaderboard dataset 62%, Verifiable Visual Rewards Tra… 27%, Rapidata/text-to-image-human-… 7% and 1 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 LMArena leaderboard dataset | 2 | 62% |
 Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts (VVRBench), arXiv:2609.35641v1 | 1 | 27% |
 Rapidata/text-to-image-human-preference-6M (Benchmark.ai Image Benchmark) published dataset-card rankings | 1 | 7% |
 GenExam leaderboard (OpenGVLab/GenExam README, relaxed score) | 1 | 3% |
| prior · pull toward the middle | 1% | |
| weighted average | = 92.1 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json