TraderBench - Analytical Reasoning: leaderboard
Metric: Analytical Reasoning score (0-100) on 9 synthetic CFA-level financial computation problems (LLM rubric: methodology 30%, calculation 30%, final answer 40%), in the authors' two-agent setup (an A2A Candidate Agent with the same six financial MCP servers for every model; GPT-5.2 judge at temperature 0 for rubric-scored sections); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro | 94.8 | #77 |
| 2 | Grok 4.1 Fast | 87.9 | #208 |
| 3 | GPT-5.2 | 87.3 | #105 |
| 4 | Qwen 3 32B (Non-reasoning) | 84 | #424 (Qwen 3 32B) |
| 5 | GLM-4.7 Flash | 82.1 | #496 |
| 6 | GPT-4o | 80.4 | #333 |
| 7 | Qwen 3 8B | 77 | #667 |
| 8 | GPT-OSS-120B | 74.8 | #330 |
| 9 | Qwen 3 30B A3B | 74.1 | #488 |
| 10 | GPT-OSS-20B | 72.1 | #499 |
| 11 | Kimi K2.5 | 71 | #139 |
| 12 | Gemma 3 27B | 68.7 | #596 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=traderbench-analytical-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-11.