Tools
Compare models
Add up to four models via ?models=slug-a,slug-b — or pick from any model page.
| Spec | DeepSeek | OpenAI |
|---|---|---|
| Params | 1.6T / 37B | 120B |
| Architecture | moe | moe |
| Context | 1M | 131K |
| Weights | ||
| License | MIT License | Apache License 2.0 |
| 1008 GBest | 75.9 GBest | |
| Cheapest hosted (out/1M) | $0.87 · OpenRouter | $0.17 · OpenRouter |
| Released | Apr 22, 2026 | Aug 4, 2025 |
| LiveBench | 71.6 | — |
| LiveBench Agentic Coding | 42.6 | — |
| LiveBench Coding | 70 | — |
| LiveBench Data Analysis | 74.5 | — |
| LiveBench Instruction Following | 62.4 | — |
| LiveBench Language | 78.1 | — |
| LiveBench Mathematics | 90.7 | — |
| LiveBench Reasoning | 82.7 | — |
| Arena Agent (IPS) | −0.008 | — |
| Arena Coding | 1502.2 | 1390.1 |
| Arena Creative Writing | 1446.1 | 1277.5 |
| Arena Hard Prompts | 1480.9 | 1362 |
| Arena Instruction Following | 1454.7 | 1325.3 |
| Arena Maths | 1445.3 | 1381.3 |
| Arena Text (overall) | 1457.9 | 1352.3 |
| Arena Code (WebDev) | 1446 | — |
| SWE-bench Verified | — | 26via mini-SWE-agent |
