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 | 291B / 37B | 1.1T / 32B |
| Architecture | moe | moe |
| Context | 1M | 262K |
| Weights | ||
| License | MIT License | Custom licence |
| 183.4 GBest | 667.5 GBest | |
| Cheapest hosted (out/1M) | $0.28 · Novita AI | $2.48 · OpenRouter |
| Released | Apr 22, 2026 | Apr 14, 2026 |
| LiveBench | 65.5 | 70.5via Thinking |
| LiveBench Agentic Coding | 37.6 | 46.9via Thinking |
| LiveBench Coding | 69.2 | 78.6via Thinking |
| LiveBench Data Analysis | 68 | 65.1via Thinking |
| LiveBench Instruction Following | 63.1 | 64.4via Thinking |
| LiveBench Language | 70.1 | 75.1via Thinking |
| LiveBench Mathematics | 79.7 | 84.3via Thinking |
| LiveBench Reasoning | 70.6 | 79.4via Thinking |
| Arena Agent (IPS) | −0.03 | −0.011 |
| Arena Coding | 1483.5 | 1514.8 |
| Arena Creative Writing | 1407.4 | 1429.5 |
| Arena Hard Prompts | 1459.3 | 1484.8 |
| Arena Instruction Following | 1428.3 | 1454.2 |
| Arena Maths | 1426.6 | 1479.3 |
| Arena Text (overall) | 1435.6 | 1460.7 |
| Arena Code (WebDev) | 1576.5via High | 1509.5 |
