Tools
Compare models
Add up to four models via ?models=slug-a,slug-b — or pick from any model page.
Scroll sideways to see every model →
| Spec | Anthropic | Z.ai |
|---|---|---|
| Params | — | 321B |
| Architecture | — | dense |
| Context | 1M | 1.3M |
| License | ||
| License terms | — | MIT License |
| — | 202.6 GBest | |
| Cheapest hosted (out/1M) | $25.00 · Anthropic | $0.50 · Relace |
| Released | Apr 16, 2026 | Aug 25, 2026 |
| LiveBench | 76.53 | 71.59 |
| LiveBench Agentic Coding | 50.66 | 56.77 |
| LiveBench Coding | 82.09 | 78.95 |
| LiveBench Data Analysis | 78.26 | 76.4 |
| LiveBench Instruction Following | 66.74 | 52.82 |
| LiveBench Language | 77.91 | 77.31 |
| LiveBench Mathematics | 92.85 | 81.24 |
| LiveBench Reasoning | 87.19 | 77.64 |
| Arena Agent | 7th of 55 | 27th of 55 |
| Arena Agent · Recoverygets back on track after a command fails | 7th of 55 | 40th of 55 |
| Arena Agent · Steerabilitydoes what it was asked, and changes course when told | 7th of 55 | 27th of 55 |
| Arena Agent · Task outcomefinishes what the session set out to do | 21st of 55 | 11th of 55 |
| Arena Agent · Tool usereaches for the right one, and does not invent one | 6th of 55 | 9th of 55 |
| Arena Coding | 1547 | 1523 |
| Arena Creative Writing | 1484 | 1433 |
| Arena Hard Prompts | 1520 | 1499 |
| Arena Instruction Following | 1494 | 1474 |
| Arena Maths | 1492 | 1501 |
| Arena Text (overall) | 1495 | 1474 |
| Arena Code (WebDev) | 1556 | 1613 |
