← comparisons / claude-opus-5-vs-deepseek-v4-pro-0813
Claude Opus 5 vs DeepSeek V4 Pro 0813
01Head to head
| Claude Opus 5 | DeepSeek V4 Pro 0813 | |
|---|---|---|
| Creator | Anthropic | DeepSeek |
| Released | 2026-07-24 | 2026-08-12 |
| Context | 1M | 1M |
| Max output | 128K | 384K |
| Input $/M | — | — |
| Output $/M | — | — |
| Knowledge cutoff | 2026-05 | — |
| Modalities | text + image → text | text → text |
| Capabilities | tool-use, extended-thinking | tool-use, extended-thinking, structured-output |
| Providers | Anthropic API | DeepSeek API |
02Benchmarks · vendor-reported unless verified
| Claude Opus 5 | DeepSeek V4 Pro 0813 | |
|---|---|---|
| DeepSWE | 68.8% | — |
| FrontierCode (mean@5) | 53.4 | — |
| SWE-bench Multilingual | 89.5% | — |
| SWE-bench Multimodal | 59.4% | — |
| SWE-bench Pro | 79.2% | — |
| SWE-bench Verified (resolved) | 96 | — |