Meta
availableShows if the model has enough results for an index.Muse Spark 1.2
Muse Spark 1.2 is a reasoning model from Meta in the Muse Spark family. 32 benchmarks count toward its score, in 7 categories.
IndexOverall score. 50 is the middle.69.2 ±3.4
CoverageShare of the index weight with results.95%
SpeedOutput tokens per second.126/s
Input / 1MUS dollars per 1M input tokens.$1.25
Output / 1MUS dollars per 1M output tokens.$4.25
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1489 (#11)
50 is the middle of the board. The range shows the doubt in the index.
3,227 votes. Elo shows what people prefer. It does not change the score.
CapabilitiesScore per category. 50 is the middle.
50 is the middleResults
32 counted| BenchmarkThe test name. | CategoryThe capability that the test measures. | ResultThe score from the publisher. | IndexThis result on the index scale. | RunThe settings of the run. | DateDate of the result. | Published byThe source of the result. |
|---|---|---|---|---|---|---|
| LiveBench Mathematics | Math | 91.2% | 69.8 | xhigh effort | 25 Jun 2026 | LiveBench |
| Artificial Analysis GPQA Diamond | Knowledge | 90.4% | 61.2 | — | — | Artificial Analysis |
| LiveBench Reasoning | Reasoning | 90.0% | 81.0 | xhigh effort | 25 Jun 2026 | LiveBench |
| MMLU Pro | Knowledge | 88.3% | 59.6 | — | 1 Sept 2026 | Vals AI |
| VulcanBench v3 | Coding | 87.0% | 72.7 | — | — | VulcanBench contributors |
| SWE-bench | Coding | 86.6% | 67.0 | — | 1 Sept 2026 | Vals AI |
| MMMU Pro | Multimodal | 86.1% | 67.5 | — | 1 Sept 2026 | Vals AI |
| Terminal-Bench 2.1 (provider run) | Agentic | 82.9% | 73.0 | — | — | DeepSeek-AI |
| Terminal-Bench 2.1 (provider run) | Agentic | 82.9% | 73.0 | — | — | DeepSeek-AI |
| Vibe Code Bench v1.1 | Coding | 79.1% | 75.1 | OpenHands | 21 Sept 2026 | Vals AI |
| Artificial Analysis Long Context Reasoning | Reasoning | 79.0% | 62.9 | — | — | Artificial Analysis |
| LiveBench Language | Knowledge | 78.6% | 65.4 | xhigh effort | 25 Jun 2026 | LiveBench |
| LiveBench Coding | Coding | 77.5% | 66.4 | xhigh effort | 25 Jun 2026 | LiveBench |
| LiveBench Data Analysis | Reasoning | 76.5% | 62.2 | xhigh effort | 25 Jun 2026 | LiveBench |
| LiveBench Instruction Following | Instruction | 74.3% | 77.3 | xhigh effort | 25 Jun 2026 | LiveBench |
| Artificial Analysis Coding Index | Coding | 72.2% | 69.8 | — | — | Artificial Analysis |
| Terminal-Bench 2.1 | Agentic | 69.7% | 65.2 | — | 21 Sept 2026 | Vals AI |
| SimpleQA Verified | Knowledge | 60.3% | 77.4 | xhigh effort | — | Epoch AI |
| DeepSWE | Agentic | 59.3% | 66.9 | — | — | Datacurve AI |
| LiveBench Agentic Coding | Agentic | 57.6% | 70.9 | xhigh effort | 25 Jun 2026 | LiveBench |
| Artificial Analysis SciCode | Coding | 57.4% | 72.5 | — | — | Artificial Analysis |
| SkillsBench | Coding | 53.0% | 68.0 | OpenHands | 11 Sept 2026 | Vals AI |
| IOI v1 | Coding | 49.5% | 68.1 | — | 9 Aug 2026 | Vals AI |
| GDPval-AA normalized | Agentic | 49.1% | 73.3 | — | — | Artificial Analysis |
| Artificial Analysis Humanity's Last Exam | Knowledge | 45.5% | 73.9 | — | — | Artificial Analysis |
| Artificial Analysis Omniscience Accuracy | Knowledge | 45.4% | 70.0 | — | — | Artificial Analysis |
| Artificial Analysis Agentic Index | Agentic | 44.0% | 73.0 | — | — | Artificial Analysis |
| ProofBench v1.1 | Math | 43.0% | 66.8 | — | 21 Sept 2026 | Vals AI |
| Artificial Analysis Intelligence Index | Knowledge | 39.6% | 71.9 | — | — | Artificial Analysis |
| Code Migration | Coding | 30.0% | 64.1 | — | 21 Sept 2026 | Vals AI |
| IOI | Coding | 21.8% | 55.1 | — | 21 Sept 2026 | Vals AI |
| Critical Physics Tasks | Reasoning | 17.7% | 75.5 | — | — | Artificial Analysis |
| FrontierSWE v2 | Coding | 12.0% | 61.2 | — | — | Proximal |
| Agent Arena command recovery | Agentic | 7.3 | 75.6 | xhigh effort | 15 Sept 2026 | LMArena |
| Terminal-Bench 4.0 | Agentic | 5.6% | 63.4 | — | 21 Sept 2026 | Vals AI |
| Agent Arena task outcome | Agentic | -2.0 | 65.2 | xhigh effort | 15 Sept 2026 | LMArena |
| Agent Arena steerability | Agentic | -4.6 | 62.2 | xhigh effort | 15 Sept 2026 | LMArena |
32 benchmarks count, from 37 of 37 results. A grey row does not count. Too few models took that benchmark.