Compare / Muse Spark 1.1 vs Kimi K2.6
Muse Spark 1.1 vs Kimi K2.6
20 results from 5 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | Muse Spark 1.1 | Kimi K2.6 |
|---|---|---|---|
| Arena (formerly LMArena)reported | OverallArena rating | 1,479 | 1,450 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,463 | 1,456 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,493 | 1,455 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,449 | 1,433 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,504 | 1,503 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,473 | 1,456 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,491 | 1,461 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,460 | 1,441 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 71.3 | 61.8 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 33.7 | 27.0 |
| Artificial Analysisreported | GPQA Diamond% of questions | 89.8% | 91.1% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 46.2% | 37.5% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 77.7% | 81.0% |
| Artificial Analysisreported | SciCode% of problems | 58.8% | 51.5% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 77.9% | 65.9% |
| Artificial Analysisreported | Terminal-Bench 4.0% of tasks | 6.1% | 0.5% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 31.8% | 23.3% |
| GDP.pdfreported | Professional document tasks% of rubric | 15.0% | 12.0% |
| SimpleQA Verified (Epoch AI)reported | Correct answers% of questions | 57.8% | 34.9% |
| Vending-Bench 2reported | Money after a yearUS dollars | $6,520.48 | $6,204.57 |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.