BrokenArXiv 05/2026
Math · 2026-06-11
This MathArena track evaluates models on the BrokenArXiv 05/2026 problem set. Scores report the percentage of problems answered correctly.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 95.0 |
| GPT-6 Astra | 91.0 |
| GPT-6.1 Sol | 90.0 |
| GPT-6 Sol | 87.0 |
| Claude Fable 5.1 | 78.0 |
| GPT-5.5 | 50.0 |
| Claude Fable 5 | 44.5 |
| Kimi K3 | 42.5 |
| Grok 4.7 | 40.0 |
| Muse Spark 1.3 | 36.5 |
| DeepSeek V4.1 Flash | 28.5 |
| Opus 4.8 | 23.0 |
| DeepSeek-V4-Flash | 19.5 |
| Gemini 3.8 Flash | 19.0 |
| Gemini 3.1 Pro Preview | 17.0 |
| Gemini 3.5 Flash | 11.0 |
| GLM-5.2 | 10.5 |
| GLM-5.1 | 10.0 |
| Gemini 3.7 Flash | 8.5 |
| Step 3.7 Flash | 8.5 |
| Gemini 3.6 Flash | 4.0 |