TRAIL GAIA: leaderboard
TRAIL (Trace Reasoning and Agentic Issue Localization) GAIA subset: evaluates LLMs on trace-based debugging and root cause analysis in agentic systems.
Metric: Joint Accuracy. Source: huggingface.co. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 05-06) | 18.3 | #147 |
| 2 | Gemini 2.5 Flash (Preview 04-17) | 10 | #237 |
| 3 | Claude 3.7 Sonnet | 4.7 | #249 |
| 4 | GPT-4.1 | 2.8 | #248 |
| 5 | Llama 4 Scout Instruct | 0 | #539 |
| 6 | Llama 4 Maverick Instruct | 0 | #456 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=trail-gaia · How It Works · Data refreshed daily, snapshot 2026-10-20.