Meta model product
Muse Glimmer 30B
Muse Glimmer-30B is Meta Superintelligence Labs' open-weight, dense multimodal model for autonomous agentic work on consumer hardware.
Updated Aug 12, 2026. Default version: Muse Glimmer-30B
LLMBoard score58.5Muse Glimmer-30B
Coverage100%19 benchmark families
Context window131.1KTokens
Official input priceN/AOfficial price unavailable
Muse Glimmer 30B Specifications
Technical details for the model's default version.
- Version
- Muse Glimmer-30B
- Released
- Aug 10, 2026
- Knowledge cutoff
- Jan 4, 2026
- Parameters
- 29.6B
- Context window
- 131.1K
- Max output
- 131.1K
- Inputs
- image, text
- Outputs
- text
- Open weights
- Yes
- License
- Apache 2.0
Muse Glimmer 30B Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Muse Glimmer 30B Benchmark Results
Benchmark scores for Muse Glimmer-30B.
| AA-LCR | 80.0% | 01 | 16 | 100.0% | C | |
| Beam 128K | 65.1% | 01 | 1 | 100.0% | C | |
| CI Memories Coverage | 64.8% | 01 | 1 | 100.0% | C | |
| CI Memories Violation Rate | 26.4% | 01 | 1 | 100.0% | C | |
| GAIA2 | 43.3% | 01 | 1 | 100.0% | C | |
| Siren AgentDojo Attack Success Rate | 28.4% | 01 | 1 | 100.0% | C | |
| Siren AgentDojo Utility | 94.2% | 01 | 1 | 100.0% | C | |
| Tau3 Banking | 23.5% | 02 | 3 | 50.0% | C | |
| WildClawBench | 47.6% | 04 | 5 | 25.0% | C | |
| AIME 2026 | 94.7% | 05 | 18 | 76.5% | C | |
| ScreenSpot Pro | 75.4% | 06 | 25 | 79.2% | C | |
| IFBench | 77.0% | 07 | 29 | 78.6% | C | |
| SkillsBench | 44.3% | 07 | 8 | 14.3% | C | |
| DeepSearchQA | 74.6% | 09 | 9 | 0.0% | C | |
| SciCode | 43.6% | 10 | 19 | 50.0% | C | |
| MCP Atlas | 75.5% | 12 | 31 | 63.3% | C | |
| OmniDocBench 1.5 | 75.8% | 13 | 17 | 25.0% | C | |
| OSWorld-Verified | 65.9% | 17 | 23 | 27.3% | C | |
| Terminal-Bench 2.1 | 51.7% | 19 | 19 | 0.0% | C | |
| CharXiv-R | 78.8% | 22 | 48 | 55.3% | C | |
| SWE-Bench Verified | 76.0% | 34 | 105 | 68.3% | C | |
| MMMU-Pro | 74.0% | 35 | 66 | 47.7% | C | |
| SWE-Bench Pro | 51.2% | 42 | 45 | 6.8% | C | |
| Humanity's Last Exam | 22.0% | 56 | 93 | 40.2% | C | |
| GPQA | 83.5% | 65 | 234 | 72.5% | C |
Muse Glimmer 30B Arena Results
Preference and agent-evaluation results for the default version.
| webdev | overall | 76 | 1358.7 | 1,507 | N/A | |
| text style control | overall | 97 | 1426.1 | 3,735 | N/A | |
| text factuality | overall | 128 | 1403.4 | 3,735 | N/A | |
| text | overall | 142 | 1390.7 | 3,735 | N/A |
Muse Glimmer 30B Runtime Performance
Provider-specific output speed and catalog latency for Muse Glimmer-30B. Runtime does not affect the capability score.
No runtime data
Browse runtime rankingsNo provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Muse Glimmer 30B Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- N/A
- Official provider
- N/A
- Lowest third-party
- From $0.35 input, $1.5 output per 1M via Kilo Gateway
- Tracked offerings
- 3
| Kilo Gateway | meta/muse-glimmer-30b | global | $0.35 | $1.5 | 131.1K | |
| Vercel AI Gateway | meta/muse-glimmer-30b | global | $0.35 | $1.5 | 131.1K | |
| OpenRouter | meta/muse-glimmer-30b | global | $0.35 | $1.5 | 131.1K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Muse Glimmer 30B Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
| Muse Glimmer-30B | 58.5 | 29.6B | 131.1K | 131.1K | Yes | Apache 2.0 |
What is Muse Glimmer 30B?
Key information about Muse Glimmer 30B and its available data.
Muse Glimmer-30B is Meta Superintelligence Labs' open-weight, dense multimodal model for autonomous agentic work on consumer hardware. 8B-parameter ViT-G/14 perception encoder, accepts interleaved text and images, produces text, and supports a 131,072-token context window.
It combines controllable multi-step reasoning, schema-based tool use, coding, failure recovery, and compatibility with agent scaffolds such as OpenClaw and Hermes Agent. Meta also releases 4-bit variants sized for 24 GB and 32 GB systems and a DFlash speculative-decoding drafter. 0.
Data as of 2026-08-11.
Muse Glimmer 30B vs nearby models
Open a comparison with the three ranked models immediately above and below this model.
Models similar to Muse Glimmer 30B
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
ME
Muse Spark
Meta
ME
Muse Spark 1.2
Meta
ME
Muse Spark 1.1
Meta
ME
Llama 4 Maverick
Meta
ME
Llama 3.1 405B
Meta
ME
Llama 3.3 70B
Meta
FAQ
Common questions about Muse Glimmer 30B.
When was Muse Glimmer 30B released?
Muse Glimmer 30B's default version was released on Aug 10, 2026.
How much does Muse Glimmer 30B cost?
No official standard PAYG price is currently available for Muse Glimmer 30B. The lowest tracked third-party offer starts at $0.35 input and $1.5 output via Kilo Gateway.
Who created Muse Glimmer 30B?
Muse Glimmer 30B was created by Meta.
What is the context window for Muse Glimmer 30B?
The default version has a 131.1K token context window.
Is Muse Glimmer 30B open weight?
Yes. The default version is marked as open weight under Apache 2.0.
How many API providers offer Muse Glimmer 30B?
3 provider offerings are linked to the default version.
What models should I compare Muse Glimmer 30B with?
Nearby ranked alternatives include Qwen3.5 122B A10B, GPT-5.2-Codex, ERNIE 5.0.