command-r7B-12-2024: benchmark results
Cohere Command R7B December 2024 open-weight model for efficient retrieval-augmented chat. Provider: Cohere. Released 2024-12-13. Access: Open.
Aggregate Rating 1405 ± 18, rank #1207 of 1656 rated models, from 60 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AgentDrive - Scenario | 95 | Accuracy (%) | 71 |
| Enkrypt AI - Insecure Code Risk | 15.56 | Risk Score | 70.1 |
| Grip on LLMs - HonestCityBench - Outdated Information | 14 | Appropriate acknowledgement rate (%; LLM judge) | 43.3 |
| Enkrypt AI - Safety Risk | 29.2 | Risk Score | 41.8 |
| Grip on LLMs - HonestCityBench - Non-Text Modality | 15 | Appropriate acknowledgement rate (%; LLM judge) | 41.7 |
| Konkur 1404 - Experimental Sciences | 34.78 | Accuracy (%, text-only) | 40.9 |
| Enkrypt AI - Toxicity Risk | 4.45 | Risk Score | 38.4 |
| AgentDrive - Comparative | 80 | Accuracy (%) | 37 |
| Grip on LLMs - HonestCityBench - Narrow Expertise | 1 | Appropriate acknowledgement rate (%; LLM judge) | 36.7 |
| SWE-Arena | 998 | Elo Score | 34.7 |
| Enkrypt AI - Jailbreak Risk | 12.11 | Risk Score | 33.7 |
| Grip on LLMs - Dutch tinyTruthfulQA | 51 | Accuracy (%; GP-IRT estimate from 100 tinyBenchmarks items, | 31.7 |
Interactive version: theaggregate.ai/model?slug=command-r7b-12-2024 · How It Works · Data refreshed daily, snapshot 2026-10-20.