Muse Spark vs Qwen3 8
Muse Spark 
Qwen3 8 Verdict
- Muse Spark leads on quality: 65.4 vs 63.9.
- Muse Spark is stronger in agents & tools, human preference, instruction following, knowledge, maths, reasoning.
- Qwen3 8 (max) is stronger in coding, composite, long context, multimodal.
| Metric | Muse Spark | Qwen3 8 (max) |
|---|---|---|
| BenchLeader Index | 65.4 | 63.9 |
| Agents & tools score | 66.3 | 56.1 |
| Coding score | 66.0 | 71.7 |
| Composite score | 68.4 | 70.8 |
| Human preference score | 69.0 | 67.9 |
| Instruction following score | 79.6 | – |
| Knowledge score | 64.5 | 62.3 |
| Long context score | 65.5 | 65.7 |
| Maths score | 56.3 | – |
| Multimodal score | 65.9 | 67.2 |
| Reasoning score | 70.0 | 67.7 |
| Blended price $/M | – | $2.67 |
| Output speed | – | 39 tok/s |
| Time to first answer | – | 53.2 s |
| Context window | 262k | 1M |
| GPQA Diamond | 89.8% | – |
| OTIS Mock AIME | 88.9% | – |
| Humanity's Last Exam | 40.6% | – |
| SciCode | 51.5% | – |
| ProofBench | 17.0% | – |
| Epoch Capabilities Index | 152.1 | 156.6 |
| LMArena Text | 1488 | 1480 |
| LMArena Hard Prompts | 1505 | 1502 |
| LMArena Coding | 1526 | 1520 |
| LMArena WebDev | – | 1685 |
| LMArena Vision | 1306 | 1313 |
| LMArena Agent | – | 3.9 |
| LiveBench | – | 78.5% |
| LiveBench Reasoning | – | 88.2% |
| LiveBench Coding | – | 72.9% |
| LiveBench Agentic Coding | – | 64.7% |
| LiveBench Mathematics | – | 91.3% |
| LiveBench Data Analysis | – | 78.4% |
| LiveBench Language | – | 79.7% |
| AA Intelligence Index | 31.3 | 40.3 |
| IFBench | 75.9% | – |
| AA-LCR | 78.0% | 78.3% |
| MMMU-Pro | 80.5% | 82.3% |
| AA-Omniscience | 7.2 | 3.4 |
| Terminal-Bench Hard | 45.5% | – |
| GPQA Diamond (AA) | 88.4% | 92.7% |
| Humanity's Last Exam (AA) | 40.7% | 43.0% |
| SciCode (AA) | – | 53.2% |
| τ²-Bench Telecom (AA) | 91.5% | – |
| AIME (Vals) | 96.9% | – |
| LiveCodeBench | – | 87.8% |
| MMLU-Pro | 87.3% | 88.6% |
| IOI | – | 73.0% |
| LegalBench | 84.2% | 83.6% |
| CorpFin | 65.1% | 65.8% |
| TaxEval | 77.7% | 75.5% |
| Terminal-Bench 2.1 (Vals) | – | 67.4% |
| SWE-bench (Vals) | 74.4% | 85.6% |
| GPQA Diamond (Vals) | 89.7% | 93.7% |
| Vals Index | – | 51.8 |
| 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% | – |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.