All comparisons
HumanEval — 89.9 SWE-bench Verified — 49.2 Aider Polyglot — 56.9 Chatbot Arena — 1372.0 Google-Proof Q&A — 71.5
DeepSeek V4 Pro vs DeepSeek R1
Side-by-side comparison of specifications, capabilities, and pricing.
| Released | 2026-04-24 | 2025-01-20 |
| Context window | 1M | 128K |
| Parameters | 1.6T | 671B, 37B Active |
| Architecture | mixture of experts | decoder only |
| License | MIT | Open Source |
| Knowledge cutoff | — | — |
Capabilities | ||
| Vision | ||
| Multimodal | ||
| Reasoning | ||
| Function calling | ||
| Tool use | ||
| Structured Outputs | ||
| Code execution | ||
Availability | ||
| Providers | — | |
Benchmarks