| Benchmark | Gemini 3.6 Flash | Muse Spark |
|---|---|---|
| AA Agentic Index | 40.5 | 28.7 |
| AA Intelligence | 52 | 44.3 |
| AA-LCR | 79 | 77 |
| AA-Omniscience | 22.1 | 7.2 |
| ARC-AGI-2 | 2.6 | 42.5 |
| arena_elo | 1481 | 1488 |
| arena_text_factuality | 1470 | 1467 |
| Artificial Analysis Coding Index | 69.2 | 58.6 |
| coding_arena_elo | 1538 | 1540 |
| critpt | 10.6 | 11.3 |
| gdpval | 46.1 | 32.3 |
| GPQA Diamond | 92.8 | 89.5 |
| HLE | 40.8 | 58.4 |
| MMMU-Pro | 83.2 | 80.5 |
| OmniScience Accuracy | 50 | 49.6 |
| OmniScience Non-Hallucination | 44.4 | 26.8 |
| scicode | 52.7 | 51.5 |
| simpleqa_verified | 68.7 | 66.3 |
| SWE-bench Pro | 58.7 | 55 |
| TauBench V3 - Banking | 29.9 | 19.6 |
| Terminal-Bench 2.1 | 78 | 62.2 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.