TerraBench - ToolUseScore: leaderboard
Metric: ToolUseScore (%): weighted composite of instruction validity, tool-call success, tool-selection accuracy, tool-category F1, argument accuracy and workflow-order consistency against the canonical trace, on the 403 executable TerraBench tasks (Fundamentals, Simulator-Grounded and Document-Grounded Verification tracks across eight Earth-science domains), each model driving the benchmark's TerraAgent ReAct harness with its 77 scientific tools; higher is better. Source: arxiv.org. Saturation forecast: Around June 2027. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 59.22 |
| 2 | Claude Haiku 4.5 | 48.14 |
| 3 | GPT-5.5 | 45.01 |
| 4 | Gemini 3.1 Pro (Preview) | 44.93 |
| 5 | Gemma 4 26B A4B | 41.84 |
| 6 | GPT-5.4 | 41.09 |
| 7 | Qwen 3.5 35B A3B | 39.95 |
| 8 | Gemini 2.5 Flash | 36.2 |
| 9 | Qwen 3.5 9B | 31.18 |
| 10 | Qwen 3 14B | 21.12 |
| 11 | Mistral 7B Instruct (v0.3) | 18.79 |
| 12 | Gemma 4 E4B | 16.31 |
| 13 | Qwen 3 8B | 7.97 |
| 14 | Qwen 3 1.7B | 5.4 |
| 15 | InternVL3-8B | 4.6 |
Interactive version: theaggregate.ai/benchmark?slug=terrabench-toolusescore · How It Works · Data refreshed daily, snapshot 2026-09-29.