InteractiveBench Math: leaderboard
Interactive proof domain on HLE math instances.
Metric: Accuracy (%). Source: www.interactivebench.org. Source: www.interactivebench.org. Saturation forecast: Around January 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Grok 4.1 Fast | 76.9 | #217 |
| 2 | GPT-5 Mini | 73.1 | #186 |
| 3 | Gemini 3 Flash | 61.5 | #99 |
| 4 | DeepSeek V3.2 | 48.1 | #205 |
| 5 | Qwen 3 Max | 46.2 | #208 |
| 6 | Kimi K2 (Thinking) | 34.6 | #243 (Kimi K2) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=interactivebench-math · How It Works · Data refreshed daily, snapshot 2026-10-20.