Infinity-Instruct-3M-0613-Mistral-7B: benchmark results
BAAI's Mistral 7B v0.1 tuned on Infinity-Instruct 3M plus the earlier 0613 chat stage, SFT-only with no RLHF. Provider: BAAI. Released 2024-06-21. Access: Open.
Aggregate Rating 1421 ± 18, rank #1145 of 1656 rated models, from 22 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - FaQuAD NLI | 82.42 | Macro F1 (%) | 94.4 |
| Open Korean LLM - Ko-IFEval | 42.07 | Strict accuracy, prompt/instruction mean (%) | 85.5 |
| Open Korean LLM - KorNAT-Helpful | 50.2 | Normalized accuracy (%) | 84.6 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 46.1 | Exact match, flexible extract (%) | 80.5 |
| Open LLM Leaderboard - MuSR | 43.51 | Score | 74.3 |
| Open Korean LLM Leaderboard | 39.7 | Average Score (%) | 73.4 |
| Open Ko-LLM Leaderboard | 39.7 | Average (%) | 73.2 |
| Open Portuguese LLM - ASSIN2 RTE | 91.75 | Macro F1 (%) | 70 |
| Open Korean LLM - KorNAT-Harmless | 63.48 | Normalized accuracy (%) | 63.3 |
| Open LLM Leaderboard - IFEval | 53.2 | Score | 63.3 |
| Open Korean LLM - Ko-EQ-Bench | 16.98 | EQ-Bench score | 62.7 |
| AlpacaEval 2.0 | 25.5 | LC Win Rate (%) | 61.3 |
Interactive version: theaggregate.ai/model?slug=infinity-instruct-3m-0613-mistral-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.