DeepSeek R1 Distill Llama 70B: benchmark results
DeepSeek's official R1 distillation into Llama 3.3 70B Instruct, fine-tuned on R1-generated reasoning traces (January 2025). Provider: DeepSeek. Released 2025-01-20. Access: Open.
Unified ELO 1523 ± 1, rank #582 of 1392 rated models, from 296 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| French LLM Leaderboard - Average | 60.02 | Average Score (%) | 100 |
| French LLM Leaderboard - GPQA FR | 63.19 | Score (%) | 100 |
| Open CoT - LSAT Analytical Reasoning | 22.61 | CoT Gain (%) | 100 |
| Open CoT - LogiQA | 12.62 | CoT Gain (%) | 98.5 |
| French LLM Leaderboard - BAC FR | 50.71 | Score (%) | 98.1 |
| Thai LLM NLU - wisesight_thai_sentiment_seacrowd_text | 56.76 | Accuracy (%) | 97.8 |
| Open CoT Leaderboard | 15.3 | Average CoT Gain (%) | 96.2 |
| Thai LLM - Reasoning | 7.15 | Rating (0-10) | 95.1 |
| Thai LLM - Extraction | 7.2 | Rating (0-10) | 94.4 |
| Open FinLLM Reasoning - XBRL-Math | 86.67 | Accuracy (%) | 94 |
| French LLM Leaderboard - IFEval FR | 66.17 | Score (%) | 90.7 |
| Open CoT - LogiQA 2 | 14.25 | CoT Gain (%) | 89.3 |
Interactive version: theaggregate.ai/model?slug=deepseek-r1-distill-llama-70b · How It Works · Data refreshed daily, snapshot 2026-09-05.