FINESSE-Bench - Trading and Technical Analysis: leaderboard
Metric: Accuracy (%) over the trading and technical-analysis group of FINESSE-Bench: Trading_derivatives (544), Trading_TA (413) and CFTe-like Level 1 (781), the per-dataset accuracies weighted by question count; a GPT-5.2 judge marks each answer correct or incorrect against the reference; zero-shot, one fixed prompt per task type, temperature 0 where possible, reasoning configurations with medium effort where the model offers them; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K2.5 | 84.98 |
| 2 | Qwen 3.5 Plus (2026-02-15) | 84.75 |
| 3 | GPT-5.2 (Medium) | 84.24 |
| 4 | Qwen 3.5 397B A17B | 83.31 |
| 5 | Qwen 3.5 122B A10B | 82.79 |
| 6 | Claude Sonnet 4.6 (Medium) | 82.45 |
| 7 | GLM-4.7 | 82.05 |
| 8 | GLM-5 | 81.3 |
| 9 | Qwen 3.5 27B | 80.38 |
| 10 | Qwen 3 235B A22B 2507 (Thinking) | 80.27 |
| 11 | GPT-5.4 (Medium) | 79.69 |
| 12 | Qwen 3.5 Flash (02-23) | 78.94 |
| 13 | Qwen 3.5 35B A3B | 78.83 |
| 14 | Claude 3.7 Sonnet (Thinking) | 78.54 |
| 15 | DeepSeek R1 0528 | 78.48 |
Interactive version: theaggregate.ai/benchmark?slug=finesse-bench-trading-and-technical-analysis · How It Works · Data refreshed daily, snapshot 2026-10-07.