OpenCompass Math - Arithmetic (CompassBench 2411): leaderboard
OpenCompass CompassBench 2411 edition (file dated 2024-12-02): numerical calculation sub-task of the math category. CompassBench is OpenCompass's periodic evaluation of chat models on a private question set; each edition is a separate board. Score 0-100, higher is better.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Estimated already saturated. 34 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 Mini (2024-09-12) | 96 | #412 |
| 2 | Yi Lightning | 88 | #429 |
| 3 | DeepSeek V2.5 | 87.3 | #494 |
| 4 | Grok Beta | 86.7 | #497 |
| 5 | GPT-4o (2024-11-20) | 86 | #392 |
| 6 | 360gpt2-pro | 85.7 | #557 |
| 7 | GPT-4o (2024-08-06) | 84.7 | #343 |
| 8 | Mistral Large 2 (Nov) Instruct (2411) | 82 | #480 |
| 9 | TeleChat2 | 81.3 | #599 |
| 10 | GLM-4 Plus | 81 | #398 |
| 11 | Claude 3.5 Sonnet (20241022) | 79 | #302 |
| 12 | Qwen 2.5 72B Instruct | 78 | #469 |
| 13 | Llama 3.1 405B Instruct FP8 | 74 | #400 |
| 14 | Step 2 16K | 73 | #442 |
| 15 | Mistral-Small-Instruct-2409 | 71.3 | #709 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-math-arithmetic-compassbench-2411 · How It Works · Data refreshed daily, snapshot 2026-11-05.