Llama 2 70B Chat GPTQ [GPTQ]: benchmark results
TheBloke's GPTQ publication of Llama 2 70B Chat, whose original release is meta-llama/Llama-2-70b-chat-hf. Conversion publication is credited separately from the original model creator. Quantization method does not establish a single effective bit width or exact evaluated branch for this catalog row. Provider: TheBloke. Access: Open.
Aggregate Rating 1400 ± 34, rank #6090 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - HellaSwag | 84.81 | Normalized accuracy (%) (10-shot) | 68.9 |
| Open LLM Leaderboard v1 - WinoGrande | 78.69 | Accuracy (%) (5-shot) | 64.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 62.63 | Normalized accuracy (%) (25-shot) | 59 |
| Open LLM Leaderboard v1 - MMLU | 62.74 | Accuracy (%) (5-shot) | 56.6 |
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 50.98 | MC2 (%) (0-shot) | 50.2 |
| Open LLM Leaderboard v1 - GSM8K | 18.65 | Accuracy (%) (5-shot) | 39.9 |
Interactive version: theaggregate.ai/model?slug=llama-2-70b-chat-gptq-gptq · How It Works · Data refreshed daily, snapshot 2026-11-08.