Kimi K3 vs Muse Spark
Verdict
- Kimi K3 leads on quality: 66.5 vs 65.4.
- Kimi K3 is stronger in agents & tools, composite, human preference, knowledge, long context, maths.
- Muse Spark is stronger in coding, multimodal, instruction following, reasoning.
| Metric | Kimi K3 | Muse Spark |
|---|---|---|
| BenchLeader Index | 66.5 | 65.4 |
| Agents & tools score | 67.5 | 66.3 |
| Coding score | 64.9 | 66.0 |
| Composite score | 74.1 | 68.4 |
| Human preference score | 71.2 | 69.0 |
| Knowledge score | 68.1 | 64.5 |
| Long context score | 71.1 | 65.5 |
| Maths score | 78.1 | 56.3 |
| Multimodal score | 65.1 | 65.9 |
| Instruction following score | – | 79.6 |
| Reasoning score | – | 70.0 |
| Blended price $/M | $6.00 | – |
| Output speed | 37 tok/s | – |
| Time to first answer | 57.7 s | – |
| Context window | 1.0M | 262k |
| GPQA Diamond | – | 89.8% |
| OTIS Mock AIME | – | 88.9% |
| Humanity's Last Exam | – | 40.6% |
| SciCode | 58.7% | 51.5% |
| APEX-Agents | 39.3% | – |
| FrontierCode | 44.2% | – |
| ProofBench | 87.0% | 17.0% |
| Epoch Capabilities Index | 157.6 | 152.1 |
| LMArena Text | – | 1488 |
| LMArena Hard Prompts | – | 1505 |
| LMArena Coding | – | 1526 |
| LMArena Vision | – | 1306 |
| LiveBench | 79.2% | – |
| LiveBench Reasoning | 90.7% | – |
| LiveBench Coding | 81.5% | – |
| LiveBench Agentic Coding | 62.2% | – |
| LiveBench Mathematics | 84.4% | – |
| LiveBench Data Analysis | 78.7% | – |
| LiveBench Language | 85.5% | – |
| AA Intelligence Index | 43.8 | 31.3 |
| IFBench | – | 75.9% |
| AA-LCR | 88.7% | 78.0% |
| MMMU-Pro | 80.5% | 80.5% |
| AA-Omniscience | 19.7 | 7.2 |
| Terminal-Bench Hard | – | 45.5% |
| GPQA Diamond (AA) | 93.5% | 88.4% |
| Humanity's Last Exam (AA) | 46.9% | 40.7% |
| SciCode (AA) | 59.5% | – |
| τ²-Bench Telecom (AA) | – | 91.5% |
| AIME (Vals) | – | 96.9% |
| MMLU-Pro | – | 87.3% |
| LegalBench | 86.0% | 84.2% |
| CorpFin | 71.6% | 65.1% |
| TaxEval | 75.7% | 77.7% |
| Terminal-Bench 2.1 (Vals) | 80.9% | – |
| SWE-bench (Vals) | 93.4% | 74.4% |
| GPQA Diamond (Vals) | 92.9% | 89.7% |
| Vals Index | 57.8 | – |
| SWE-Bench Pro | – | 55.0% |
| MCP Atlas | – | 82.2% |
| MultiChallenge | – | 75.5% |
| PRBench Finance | – | 52.4% |
| PRBench Legal | – | 52.3% |
| MultiNRC | – | 59.0% |
| TutorBench | – | 68.5% |
| EQ-Bench 4 | 1339 | – |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.