| Benchmark | Gemini 3.5 Flash | Muse Spark |
|---|---|---|
| AA Agentic Index | 70.4 | 28.7 |
| AA Intelligence | 52 | 44.3 |
| AA-LCR | 81 | 77 |
| AA-Omniscience | 21.2 | 7.2 |
| ARC-AGI-2 | 72.1 | 42.5 |
| arena_elo | 1477 | 1488 |
| arena_text_factuality | 1466 | 1467 |
| Artificial Analysis Coding Index | 70.1 | 58.6 |
| coding_arena_elo | 1490 | 1540 |
| critpt | 13.1 | 11.3 |
| frontiermath_tier_4 | 14.6 | 14.6 |
| gdpval | 57.8 | 32.3 |
| GPQA Diamond | 92.2 | 89.5 |
| HLE | 42.7 | 58.4 |
| IFBench | 76.3 | 75.9 |
| itbenchSre | 40.3 | 16.2 |
| MCP Atlas | 83.6 | 82.2 |
| MMMU-Pro | 84.3 | 80.5 |
| OmniScience Accuracy | 51.4 | 49.6 |
| OmniScience Non-Hallucination | 38.2 | 26.8 |
| scicode | 53.1 | 51.5 |
| simpleqa_verified | 68.4 | 66.3 |
| SWE-bench Pro | 55.1 | 55 |
| TauBench V3 - Banking | 32.2 | 19.6 |
| Terminal-Bench 2.0 | 76.2 | 59 |
| Terminal-Bench 2.1 | 78.7 | 62.2 |
| Terminal-Bench Hard | 46.2 | 45.5 |
| τ²-Bench Telecom (AA run) | 95.6 | 91.5 |
| τ³-Bench | 25.4 | 19.6 |
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.