deepseek-llm-7B-chat: benchmark results
DeepSeek LLM 7B chat checkpoint. Provider: DeepSeek. Released 2023-11-29. Access: Open.
Aggregate Rating 1364 ± 18, rank #1354 of 1656 rated models, from 112 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 46.68 | Score | 90.1 |
| ChineseSafe Benchmark | 71.63 | Accuracy (%) | 79.6 |
| Open Chinese LLM - HellaSwag | 62.57 | Accuracy (%) | 67.8 |
| Open Chinese LLM - WinoGrande | 64.17 | Accuracy (%) | 64.8 |
| OpenCompass Code - HumanEval - Chinese (CompassBench 2404) | 42.1 | Score (%) | 63.3 |
| OpenCompass Code - HumanEval-X (CompassBench 2404) | 33.2 | Score (%) | 63.3 |
| OpenCompass Reasoning - Abductive - Chinese (CompassBench 2404) | 51 | Score (%) | 63.3 |
| OpenCompass Reasoning - Inductive - English (CompassBench 2404) | 29.2 | Score (%) | 60 |
| Uni-SafeBench - Malicious Text Generation | 91.7 | Both-Safe rate (%) on the 630 malicious text generation quer | 60 |
| LongPIBench - Resume Screening | 97 | Attack success rate (%; share of 100 long synthetic resumes | 57.1 |
| OpenCompass Code - MBPP - English (CompassBench 2404) | 49.8 | Score (%) | 56.7 |
| OpenCompass Agent - CIBench - English (CompassBench 2404) | 39.3 | Score (%) | 53.3 |
Interactive version: theaggregate.ai/model?slug=deepseek-llm-7b-chat · How It Works · Data refreshed daily, snapshot 2026-10-20.