Claude Fable 5 vs Muse Spark
Claude Fable 5 
Muse Spark Verdict
- Claude Fable 5 leads on quality: 70.4 vs 65.4.
- Claude Fable 5 is stronger in agents & tools, coding, composite, human preference, knowledge, long context, multimodal, reasoning.
- Muse Spark is stronger in instruction following, maths.
| Metric | Claude Fable 5 | Muse Spark |
|---|---|---|
| BenchLeader Index | 70.4 | 65.4 |
| Agents & tools score | 74.4 | 66.3 |
| Coding score | 70.6 | 66.0 |
| Composite score | 91.6 | 68.4 |
| Human preference score | 71.3 | 69.0 |
| Instruction following score | 62.6 | 79.6 |
| Knowledge score | 71.3 | 64.5 |
| Long context score | 67.8 | 65.5 |
| Multimodal score | 69.6 | 65.9 |
| Reasoning score | 75.6 | 70.0 |
| Maths score | – | 56.3 |
| Blended price $/M | $20.00 | – |
| Output speed | 63 tok/s | – |
| Time to first answer | 92.2 s | – |
| Context window | 1M | 262k |
| GPQA Diamond | – | 89.8% |
| OTIS Mock AIME | – | 88.9% |
| Humanity's Last Exam | – | 40.6% |
| SciCode | – | 51.5% |
| Remote Labor Index | 16.1% | – |
| APEX-Agents | 45.0% | – |
| FrontierCode | 53.5% | – |
| ProofBench | – | 17.0% |
| Epoch Capabilities Index | 163.4 | 152.1 |
| LMArena Text | 1507 | 1488 |
| LMArena Hard Prompts | 1533 | 1505 |
| LMArena Coding | 1551 | 1526 |
| LMArena WebDev | 1628 | – |
| LMArena Vision | 1330 | 1306 |
| AA Intelligence Index | 49.7 | 31.3 |
| IFBench | 63.5% | 75.9% |
| AA-LCR | 82.3% | 78.0% |
| MMMU-Pro | – | 80.5% |
| AA-Omniscience | 43.3 | 7.2 |
| Terminal-Bench Hard | 62.9% | 45.5% |
| GPQA Diamond (AA) | 92.6% | 88.4% |
| Humanity's Last Exam (AA) | 55.5% | 40.7% |
| SciCode (AA) | 61.0% | – |
| τ²-Bench Telecom (AA) | 98.5% | 91.5% |
| AIME (Vals) | – | 96.9% |
| LiveCodeBench | 89.8% | – |
| MMLU-Pro | 91.5% | 87.3% |
| IOI | 72.3% | – |
| LegalBench | 88.6% | 84.2% |
| CorpFin | 71.8% | 65.1% |
| TaxEval | 76.9% | 77.7% |
| Terminal-Bench 2.1 (Vals) | 80.5% | – |
| SWE-bench (Vals) | 95.0% | 74.4% |
| GPQA Diamond (Vals) | 93.2% | 89.7% |
| Vals Index | 66.0 | – |
| SWE-Bench Pro | – | 55.0% |
| MCP Atlas | 83.3% | 82.2% |
| MultiChallenge | – | 75.5% |
| PRBench Finance | 53.9% | 52.4% |
| PRBench Legal | 52.6% | 52.3% |
| MultiNRC | – | 59.0% |
| HiL-Bench | 56.3% | – |
| TutorBench | – | 68.5% |
| EQ-Bench 4 | 1340 | – |
| Kagi LLM Benchmark | 88.8% | – |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.