Across 16 shared benchmarks, Gemini 3.1 Pro scores higher on 13 and muse-glimmer-30b on 3. The widest gap is HLE, where Gemini 3.1 Pro scores 51.4 against 22. muse-glimmer-30b is the cheaper of the two on tracked API pricing ($0.30 against $2.00 per million input tokens).
| Benchmark | Gemini 3.1 Pro | muse-glimmer-30b |
|---|---|---|
| AA-LCR | 79 | 80 |
| aime_2026 | 98.3 | 94.7 |
| baby_vision | 51.6 | 70.4 |
| charxiv_rq | 89.9 | 85.9 |
| deepsearchqa_f1 | 81.9 | 74.6 |
| GPQA Diamond | 94.3 | 83.5 |
| HLE | 51.4 | 22 |
| IFBench | 77.1 | 77 |
| MCP Atlas | 78.2 | 75.5 |
| MMMU-Pro | 83 | 74 |
| OSWorld-Verified | 76.2 | 65.9 |
| scicode | 59 | 43.6 |
| SWE-bench Pro | 54.2 | 51.2 |
| SWE-bench Verified | 80.6 | 76 |
| Tau 3 Banking | 16.5 | 23.5 |
| Terminal-Bench 2.1 | 74 | 51.7 |
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. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (google-official, deepinfra), otherwise the lowest tracked offer.