Open-weight 1T MoE (42B active) matching frontier intelligence at half the cost. Strong agentic and reasoning, awaiting coding eval coverage.
| Benchmark | Score | Rank |
|---|---|---|
Arena EloArtificial Analysis Human preference ranking via blind comparisons | 1571 | #3 / 47 |
TerminalArtificial Analysis Agentic terminal coding tasks requiring multi-step execution | 43.2% | #19 / 43 |
GPQAArtificial Analysis PhD-level science questions even experts struggle with | 86.6% | #24 / 59 |