Llama 3 70B Synthia V3.5: benchmark results
Migel Tissera's uncensored general-assistant fine-tune of Llama 3 70B on the Synthia v3.5 varied-system-context dataset. Provider: Meta. Released 2024-05-26. Access: Open.
Aggregate Rating 1602 ± 41, rank #992 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Chinese LLM - CMMLU | 73.46 | Accuracy (%) | 97.6 |
| Open LLM Leaderboard - MuSR | 49.22 | Score | 97.3 |
| Open LLM Leaderboard - GPQA | 38.76 | Score | 96.9 |
| Open Chinese LLM - GSM8K | 69.14 | Accuracy (%) | 93.8 |
| Open Chinese LLM Leaderboard | 69.38 | Official Nonzero-Task Mean (%) | 93.2 |
| Open LLM Leaderboard - BBH | 64.89 | Score | 90.7 |
| Open Chinese LLM - HellaSwag | 68.46 | Accuracy (%) | 87.9 |
| Open Chinese LLM - ARC Challenge | 59.9 | Accuracy (%) | 86.8 |
| Open Chinese LLM - WinoGrande | 67.96 | Accuracy (%) | 85.5 |
| Open LLM Leaderboard - MMLU-Pro | 46.58 | Score | 85.5 |
| Open Chinese LLM - C-Eval Semantic | 87.41 | Accuracy (%) | 84.9 |
| Open LLM Leaderboard - MATH Level 5 | 21.15 | Score | 72.9 |
Interactive version: theaggregate.ai/model?slug=llama-3-70b-synthia-v3-5 · How It Works · Data refreshed daily, snapshot 2026-11-08.