OpenCompass Agent - T-Eval - English (CompassBench 2409): leaderboard
OpenCompass CompassBench 2409 edition (file dated 2024-10-09): general tool utilization (T-Eval) sub-task of the agent category, English questions. 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. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Step 2 16K | 88.3 | #442 |
| 2 | Llama 3.1 70B Instruct | 88.1 | #567 |
| 3 | DeepSeek V2.5 | 88.1 | #494 |
| 4 | GPT-4o (2024-08-06) | 88 | #343 |
| 5 | Llama 3.1 405B Instruct FP8 | 87.8 | #400 |
| 6 | Mistral Large 2 (Jul) | 87.7 | #490 |
| 7 | Qwen 2.5 72B Instruct | 86.9 | #469 |
| 8 | GPT-4o (2024-05-13) | 86.8 | #337 |
| 9 | Yi Large | 86.7 | #566 |
| 10 | GPT-4o Mini (2024-07-18) | 86.6 | #568 |
| 11 | Mistral-Small-Instruct-2409 | 86.5 | #709 |
| 12 | Gemini 1.5 Pro | 85.9 | #454 |
| 13 | GLM-4 9B Chat | 85.9 | #911 |
| 14 | Gemma 2 27B (IT) | 85.7 | #652 |
| 15 | GLM-4 Plus | 84.3 | #398 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-agent-t-eval-english-compassbench-2409 · How It Works · Data refreshed daily, snapshot 2026-11-05.