GPT Image 2.5 Tops Image Arena, but Old Prompts Cause Artifacts, User Ablations Find
altryne · x · 2026-09-12
Per Artificial Analysis' Image Arena, OpenAI's new GPT Image 2.5 (Flare/Sunburst) takes the top two spots, generating higher-quality images in less than half the time of GPT Image 2 at the same price.
User altryne re-ran thumbnail ablations and confirmed GPT Image 2.5 beats Nano Banana, his previous winner — but warns that his old prompting style made GPT images "overcook" thumbnails with artifacts, so users should adjust prior prompts before judging the new model.
Related event: OpenAI's GPT Image 2.5 Tops Image Arena with Halved Latency(3 posts)→
More from Models
- Why do Sonnet 5 and Opus 5 feel worse than GLM 5.3? Distillation's limits, dissected — baseten · 2026-09-12
- OpenAI's GPT-Rosalind exits research preview, open to orgs via API and Codex — OpenAIDevs · 2026-09-12
- Codex adds Life Sciences plugins for genomics, protein structure and translational work — OpenAIDevs · 2026-09-12
- OpenAI launches GPT-Rosalind for biological reasoning in API and Codex — OpenAIDevs · 2026-09-12
- DeepSeek's edge: each model iteration targets the last one's biggest bottleneck — xennygrimmato_ · 2026-09-12
- RL on just 1,700 tasks lifts Kimi K2.7 across five coding benchmarks — echen · 2026-09-12