Muse Glimmer is a 30-billion-parameter open-weight multimodal model from Meta Superintelligence Labs, distilled from Muse Spark for always-on local agents, tool use, coding, and image understanding.
Muse Glimmer 30B by meta is offered by 14 providers on this page. Public prices are shown for 12 of them. Its reference price is $0.35 per 1M input tokens and $1.50 per 1M output tokens. The lowest paid channel is EmpirioLabs AI at $0.20 / $0.80 per 1M, about 1.7× below the reference price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it. It also has 2 free ($0) channels; free tiers usually carry rate limits, and subscription-covered access bills $0 per token only after the subscription fee.
The context window is 131,072 tokens, with an output limit of 131,072 tokens. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, PDF, and Video. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Its training knowledge cuts off at 2026-01-04.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| Nvidia | Cloud | Free | — | — | 131,072 | 131,072 | ||
| Requesty | Gateway | Free | — | — | 131,072 | 20,480 | ||
| EmpirioLabs AI | Gateway | $0.20 | $0.80 | $0.05 | — | 131,072 | 32,768 | |
| Vultr | Cloud | $0.25 | $1.00 | — | — | 131,072 | 131,072 | |
| Kilo Gateway | Gateway | $0.30 | $1.10 | $0.04 | — | 131,072 | 117,964 | |
| DevPass (LLM Gateway) | Gateway | $0.30 | $1.20 | $0.04 | — | 131,072 | 131,072 | |
| LLM Gateway | Gateway | $0.30 | $1.20 | $0.04 | — | 131,072 | 16,384 | |
| Eden AI deepinfra/meta-models/Muse-Glimmer-30B | Gateway | $0.30 | $1.20 | $0.04 | — | 131,072 | 131,072 | |
| NanoGPT | Gateway | $0.35 | $1.50 | $0.04 | — | 131,072 | 117,964 | |
| OpenRouter | Gateway | $0.35 | $1.50 | $0.04 | — | 131,072 | 117,964 | |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 1 of 14 providers.
2 more channels offer $0 (Nvidia, Requesty); free tiers usually have rate limits and no SLA, excluded from ranking.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| MCP Atlas | variant: public | 75.5 | success rate | Source ↗ |
| DeepSearch QA | — | 74.6 | — | Source ↗ |
| SWE-Bench Pro | — | 51.2 | resolve rate | Source ↗ |
| SWE-Bench Verified | — | 76 | resolve rate | Source ↗ |
| Terminal-Bench | variant: with terminus2 · v2.1 | 51.7 | success rate | Source ↗ |
| OSWorld-Verified | — | 65.9 | success rate | Source ↗ |
| AIME 2026 | — | 94.7 | accuracy | Source ↗ |
| GPQA Diamond | variant: AA | 83.5 | accuracy | Source ↗ |
| CharXiv Reasoning | — | 78.8 | accuracy | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.