FINESSE-Bench - CFTe-like Level 1: leaderboard
Metric: Accuracy (%) on the 781 CFTe-like Level 1 multiple-choice questions on basic technical analysis (chart types, trends, support and resistance, moving averages) of FINESSE-Bench; 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 | Qwen 3.5 122B A10B | 86.81 |
| 2 | Qwen 3.5 Plus (2026-02-15) | 86.81 |
| 3 | Qwen 3.5 397B A17B | 85.53 |
| 4 | Kimi K2.5 | 84.76 |
| 5 | Claude Sonnet 4.6 (Medium) | 84.64 |
| 6 | Qwen 3.5 27B | 84.38 |
| 7 | GLM-4.7 | 84.25 |
| 8 | Qwen 3 235B A22B 2507 (Thinking) | 83.99 |
| 9 | GPT-5.2 (Medium) | 83.99 |
| 10 | Qwen 3.5 Flash (02-23) | 83.1 |
| 11 | Claude 3.7 Sonnet (Thinking) | 82.59 |
| 12 | Llama 4 Maverick | 82.2 |
| 13 | GPT-5.4 (Medium) | 82.2 |
| 14 | Qwen 3.5 35B A3B | 81.18 |
| 15 | GLM-5 | 81.18 |
Interactive version: theaggregate.ai/benchmark?slug=finesse-bench-cfte-like-level-1 · How It Works · Data refreshed daily, snapshot 2026-10-07.