Models · Compare
DeepSeek-Coder-V2 Instruct vs Claude 3.5 Sonnet
Rows highlighted in warm gray are where the models differ. Numbers carry their as-of date and primary source.
Specs
| Field | A: DeepSeek-Coder-V2 Instruct | B: Claude 3.5 Sonnet |
|---|---|---|
| Released | 2024-06-17 | 2024-06-20 |
| Developer | DeepSeek | Anthropic |
| Openness | Open weights | Proprietary |
| License | DeepSeek License | Proprietary |
| OSI-approved | no | no |
| Data released | no | no |
| Training code | no | no |
| Architecture | moe | unknown |
| Total params | 236B | — |
| Active params | 21B | — |
| Experts | — | — |
| Context window | 128K | 200K |
| Attention | mla | unknown |
| Position enc. | rope-yarn | unknown |
| Pretraining tokens | 6.0T | — |
| Post-training | sft, rlhf | rlhf, constitutional |
| Training hardware | — | — |
| $/M input | — | — |
| $/M output | — | — |
| Output tok/sec | — | — |
Benchmarks
Missing scores render as not reported; never inferred. Bold highlights the leader per benchmark.
General reasoning
Code
| HumanEval | 90.2 2024-06-17 | — |
| LiveCodeBench | 43.4 2024-06-17 | — |
Context · A
Code specialist branch of the V2 line. Continued pretraining from an intermediate V2 checkpoint on an additional 6 trillion code and math tokens, expanding programming language coverage from 86 to 338 and extending context from 16K to 128K. The Instruct variant reported HumanEval scores competitive with GPT-4 Turbo.
Context · B
The first Claude release to beat its own larger sibling (Claude 3 Opus) on most benchmarks. Established Artifacts, driving a wave of code-and-canvas product copies.
DeepSeek-Coder-V2 Instruct detail → · Claude 3.5 Sonnet detail →