Model
DeepSeek Chat
DeepSeek · open weights
Knowledge & Reasoning
Knowledge & Reasoning| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| GPQA Diamond | 71.2 | accuracy | N/A | Independent | high | epoch.ai |
| MS-Exam-Gen | 86.9 | accuracy | N/A | Independent | medium | arxiv.org |
Coding
Coding| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| Aider Polyglot | 70.2 | percent_correct | N/A | Independent | medium | aider.chat |
| claimcheck | 81.3 | macro_recall | N/A | Independent | medium | arxiv.org |
| PolyCodeEval | 65.9 | pass_rate | repocoder | Independent | medium | arxiv.org |
| PolyCodeEval | 56.0 | pass_rate | hcp | Independent | medium | arxiv.org |
| PolyCodeEval | 50.7 | pass_rate | llm direct | Independent | medium | arxiv.org |
| PolyCodeEval | 36.7 | pass_rate | aligncoder | Independent | medium | arxiv.org |
Agentic / Tool Use
Agentic / Tool Use| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| Grounding Awareness | 0.4 | sensitivity | gl | Independent | medium | arxiv.org |
Multimodal
MultimodalInstruction Following
Instruction FollowingSafety & Alignment
Safety & Alignment| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| MemCut | 98.2 | residual_recovery | turn rag | Independent | medium | arxiv.org |
| MemCut | 93.0 | residual_recovery | memorybank | Independent | medium | arxiv.org |
| MemCut | 88.8 | residual_recovery | lightmem | Independent | medium | arxiv.org |
| MemCut | 85.1 | residual_recovery | nemori | Independent | medium | arxiv.org |
| OMIT | 0.2 | obr | N/A | Independent | medium | arxiv.org |
| RT-Safe | 0.0 | safe_success_rate | N/A | Independent | medium | arxiv.org |
| Sense and Sensitivity | 78.9 | accuracy | N/A | Independent | medium | arxiv.org |
Multilingual
MultilingualReproduced by the community
Loading community runs…