Tools
Compare models
Add up to four models via ?models=slug-a,slug-b — or pick from any model page.
| Spec | DeepSeek | Moonshot AI |
|---|---|---|
| Params | 1.6T / 37B | 1.1T / 32B |
| Architecture | moe | moe |
| Context | 1M | 262K |
| Weights | ||
| License | MIT License | Custom licence |
| 1008 GBest | 667.5 GBest | |
| Cheapest hosted (out/1M) | $0.87 · OpenRouter | $2.48 · OpenRouter |
| Released | Apr 22, 2026 | Apr 14, 2026 |
| LiveBench | 71.6 | 70.5via Thinking |
| LiveBench Agentic Coding | 42.6 | 46.9via Thinking |
| LiveBench Coding | 70 | 78.6via Thinking |
| LiveBench Data Analysis | 74.5 | 65.1via Thinking |
| LiveBench Instruction Following | 62.4 | 64.4via Thinking |
| LiveBench Language | 78.1 | 75.1via Thinking |
| LiveBench Mathematics | 90.7 | 84.3via Thinking |
| LiveBench Reasoning | 82.7 | 79.4via Thinking |
| Arena Agent (IPS) | −0.008 | −0.011 |
| Arena Coding | 1502.2 | 1514.8 |
| Arena Creative Writing | 1446.1 | 1429.5 |
| Arena Hard Prompts | 1480.9 | 1484.8 |
| Arena Instruction Following | 1454.7 | 1454.2 |
| Arena Maths | 1445.3 | 1479.3 |
| Arena Text (overall) | 1457.9 | 1460.7 |
| Arena Code (WebDev) | 1446 | 1509.5 |
