Across 16 shared benchmarks, muse-glimmer-30b scores higher on 2 and Qwen3.7 Plus Preview on 12, with 2 level. The widest gap is HLE, where Qwen3.7 Plus Preview scores 35.6 against 22. muse-glimmer-30b is the cheaper of the two on tracked API pricing ($0.30 against $0.40 per million input tokens).
| Benchmark | muse-glimmer-30b | Qwen3.7 Plus Preview |
|---|---|---|
| AA-LCR | 80 | 69 |
| baby_vision | 70.4 | 70.4 |
| charxiv_rq | 85.9 | 85.9 |
| GPQA Diamond | 83.5 | 90.3 |
| HLE | 22 | 35.6 |
| IFBench | 77 | 79.1 |
| MCP Atlas | 75.5 | 73.2 |
| MMMU-Pro | 74 | 80.5 |
| OmniDocBench 1.5 | 75.8 | 91.4 |
| OSWorld-Verified | 65.9 | 73.3 |
| scicode | 43.6 | 51.3 |
| screenspot_pro_no_tools | 75.4 | 79 |
| SWE-bench Pro | 51.2 | 57.6 |
| SWE-bench Verified | 76 | 77.7 |
| Terminal Bench 2.1 (Terminus) | 51.7 | 64 |
| Terminal-Bench 2.1 | 51.7 | 61 |
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 (deepinfra, alibaba-official), otherwise the lowest tracked offer.