GTA: leaderboard
General Tool Agents benchmark from OpenCompass. Tests LLMs on multi-step tool use across perception, operation, logic, and creativity categories using real-world tool chains.
Metric: Answer Accuracy (%). Source: github.com. Saturation forecast: Around 2029. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 45.07 | #98 |
| 2 | DeepSeek V3.2 | 42.68 | #205 |
| 3 | Qwen 3 235B A22B | 42.21 | #313 |
| 4 | Gemini 2.5 Pro | 39.58 | #152 |
| 5 | Qwen 3 8B | 27.1 | #674 |
| 6 | Llama 3.2 3B Instruct | 24.99 | #1352 |
| 7 | Llama 4 Scout | 22.03 | #656 |
| 8 | Kimi K2 | 18.32 | #243 |
| 9 | Claude Sonnet 4.5 | 17.89 | #142 |
| 10 | Grok 4 | 17.78 | #181 |
| 11 | Qwen 3 30B A3B | 15.67 | #493 |
| 12 | Llama 3.1 70B Instruct | 13.24 | #561 |
| 13 | Llama 3.1 8B Instruct | 8.78 | #1035 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=gta · How It Works · Data refreshed daily, snapshot 2026-10-20.