DeepSeek-67B-Chat: benchmark results
Provider: DeepSeek. Access: Open.
Aggregate Rating 1468 ± 20, rank #2901 of 10916 rated models, from 73 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Code - HumanEval - Chinese (CompassBench 2404) | 63.4 | Score (%) | 88.3 |
| OpenCompass Language - Content Critique - Chinese (CompassBench 2404) | 50 | Score (%) | 88.3 |
| OpenCompass Code - HumanEval-X (CompassBench 2404) | 42.4 | Score (%) | 83.3 |
| OpenCompass Code - MBPP - English (CompassBench 2404) | 68.1 | Score (%) | 83.3 |
| OpenCompass Knowledge - Humanities - Chinese (CompassBench 2404) | 49 | Score (%) | 80 |
| OpenCompass Agent - CIBench - Chinese (CompassBench 2402) | 48.6 | Score (%) | 79.4 |
| OpenCompass Agent - CIBench - English (CompassBench 2402) | 54.5 | Score (%) | 79.4 |
| OpenCompass Agent - CIBench - English (CompassBench 2404) | 45.1 | Score (%) | 76.7 |
| OpenCompass Knowledge - Social Science - Chinese (CompassBench 2404) | 61 | Score (%) | 76.7 |
| OpenCompass LLM - Knowledge (CompassBench 2404) | 63 | Score (%) | 76.7 |
| OpenCompass LLM - Knowledge - Chinese (CompassBench 2404) | 53.7 | Score (%) | 76.7 |
| OpenCompass LLM - Agent - English (CompassBench 2402) | 59.6 | Score (%) | 73.5 |
Interactive version: theaggregate.ai/model?slug=deepseek-67b-chat · How It Works · Data refreshed daily, snapshot 2026-11-08.