Fino1-8B: benchmark results
TheFinAI's financial-reasoning fine-tune of Llama 3.1 8B Instruct, trained on FinQA-derived reasoning paths. Provider: The Fin AI. Released 2025-02-12.
Aggregate Rating 1381 ± 47, rank #6814 of 10916 rated models, from 18 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| StakeBench - Revealed Side | 52.4 | Strict accuracy (%) on G2: which side (YES or NO) a position | 92.9 |
| Open FinLLM Reasoning | 59.95 | Accuracy (%) | 84 |
| Open FinLLM Reasoning - FinQA | 73.03 | Accuracy (%) | 84 |
| Open FinLLM Reasoning - XBRL-Math | 84.44 | Accuracy (%) | 80 |
| Open FinLLM Reasoning - DM-Simplong | 56 | Accuracy (%) | 76 |
| Open FinLLM Reasoning - DM-Complong | 26.33 | Accuracy (%) | 52 |
| StakeBench - Commitment Detection | 60.1 | Accuracy (%) on G1: does a comment come from a user holding | 28.6 |
| StakeBench | 7.2 | Agg (self-reported) | 14.3 |
| FINESSE-Bench - CFA-like Level 2 | 41.98 | Accuracy (%) on the 293 CFA-like Level 2 multiple-choice ite | 6.7 |
| FINESSE-Bench - CFA-like Level 3 | 42.45 | Accuracy (%) on the 318 CFA-like Level 3 multiple-choice que | 6.7 |
| FINESSE-Bench - CFA-like Level 1 | 52.29 | Accuracy (%) on the 1069 CFA-like Level 1 multiple-choice qu | 3.3 |
| FINESSE-Bench | 46.61 | exam-like (self-reported) | 0 |
Interactive version: theaggregate.ai/model?slug=fino1-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.