Muse Spark: benchmark results
Meta's Muse Spark model, a closed high-capability system tracked among the top ranked models. Provider: Meta. Released 2026-04-08. Access: Private.
Aggregate Rating 1802 ± 15, rank #64 of 1656 rated models, from 147 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLM Stats (FrontierScience Research) | 38.3 | Score (%) | 100 |
| LLM Stats (HealthBench Hard) | 42.8 | Score (%) | 100 |
| SEAL - TutorBench | 68.55 | Score | 100 |
| TaxEval v2 | 77.68 | Score (%) | 99.2 |
| Chatbot Arena (Text - French) | 1521 | Arena Score | 99 |
| Chatbot Arena (Text - German) | 1511 | Arena Score | 99 |
| Chatbot Arena (Vision - Humor) | 1323 | Arena Score | 98.9 |
| Vals AI TaxEval v2 | 77.68 | Accuracy (%) | 98.6 |
| AIME | 96.88 | Score (%) | 98.4 |
| Vals AI AIME | 96.88 | Accuracy (%) | 98.4 |
| LLM Stats (CharXiv-R) | 86.4 | Score (%) | 97.6 |
| Chatbot Arena (Text - English) | 1496 | Arena Score | 97.5 |
Interactive version: theaggregate.ai/model?slug=muse-spark · How It Works · Data refreshed daily, snapshot 2026-10-20.