Capability
52.4/100
field median 58
#116 of 224 ranked models
Muse Glimmer 30B
Released Aug 10, 2026 — see all recent releases
Decision reading
Muse Glimmer 30B scores 52.4 out of 100 and ranks #116 of 224. This profile shows 14 source-displayable benchmark rows; its strongest eligible category is Instruction Following at #21. Published weights can be self-hosted, but infrastructure cost varies and is not a comparable API token rate.
Data as of August 22, 2026 · How the score is built
Strongest published evidence
Instruction Following ranks #21. A well-rounded choice across a range of tasks.
Validate before choosing
14 published rows leave some tracked benchmark slots empty. No comparable first-party API token rate is published.
Decision snapshot
Each value carries a field reference instead of floating alone. Markers compare this model with the current ranked and priced catalog; they are not absolute quality thresholds.
Price
Self-hosted; infrastructure cost varies
input median $1
No comparable first-party hosted token rate
Speed
Not measured
field median 91.5 tok/s
Time to first token not measured
Context
131Ktokens
field median 200,000
Reported for this model; direct source link not stored
Capability shape
Each axis shows percentile within that category’s eligible cohort. The comparison outline is the median of the six nearest public-score peers; a collapsed vertex means the category is not rank-eligible.
Muse Glimmer 30B category percentile values
- Agentic62nd percentile
- Coding56th percentile
- ReasoningNot eligible
- KnowledgeNot eligible
- MathNot eligible
- MultilingualNot eligible
- Multimodal21st percentile
- Instruction following50th percentile
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#53/137
- Coding#62/141
- ReasoningNot ranked
- KnowledgeNot ranked
- MathNot ranked
- MultilingualNot ranked
- Multimodal#27/34
- Inst. Following#21/41
Top decileTop quartileMid-fieldNot eligible
How much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic4/4 verified
- Coding4/4 verified
- ReasoningNot measured
- KnowledgeNot measured
- Math1/1 verified
- MultilingualNot measured
- Multimodal4/4 verified
- Inst. Following1/1 verified
Verified sourceProvisionalNot measured
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- Not published
- Context window
- 131K
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- Not sourced yet
- Output modalities
- Not sourced yet
- Parameters
- Not sourced yet
- Availability
- Meta publishes the full-precision and two 4-bit Muse Glimmer 30B checkpoints under Apache 2.0 on Hugging Face. The K-Quant-17GB release targets 24 GB hardware, while K-Quant-Dynamic targets 32 GB; Meta also ships a DFlash speculative-decoding drafter and a dedicated perception encoder.
- Cloud regions
- Not tracked yet
- Lifecycle
- Current
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Not documented in the pricing record
- Self-host
- Open weights available; hardware estimate not sourced
- Rate limits
- Not tracked yet
Deployment options
Self-host and provider-specific paths stay separate from benchmark evidence so operating constraints are visible before a score becomes the whole decision.
Published weights are available, but BenchLM does not yet have a sourced parameter and VRAM profile for this exact model. Hardware cost estimates stay unavailable until that sizing record is complete.
Estimate VRAM from known parametersCategory score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank #53 of 137Percentile 62ndWeight 22%4 benchmarksVerified | 51.7 | #53 of 137 | 62nd | 22% | 4 benchmarks | Verified |
| CodingRank #62 of 141Percentile 56thWeight 20%4 benchmarksVerified | 52.0 | #62 of 141 | 56th | 20% | 4 benchmarks | Verified |
| ReasoningWeight 17%0 benchmarksNot measured | Not measured | Not ranked | Not available | 17% | 0 benchmarks | Not measured |
| KnowledgeWeight 12%0 benchmarksNot measured | Not measured | Not ranked | Not available | 12% | 0 benchmarks | Not measured |
| MathRank Not rankedWeight 5%1 benchmarkVerified | 75.5 | Not ranked | Not available | 5% | 1 benchmark | Verified |
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | Not ranked | Not available | 7% | 0 benchmarks | Not measured |
| MultimodalRank #27 of 34Percentile 21stWeight 12%4 benchmarksVerified | 48.9 | #27 of 34 | 21st | 12% | 4 benchmarks | Verified |
| Inst. FollowingRank #21 of 41Percentile 50thWeight 5%1 benchmarkVerified | 84.5 | #21 of 41 | 50th | 5% | 1 benchmark | Verified |
Benchmark ledger
Coding opens by default. The marker compares each value with the best source-verified result in the catalog; provisional leaders do not set the reference. Expand the remaining categories for every published row.
Coding4 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| SWE-bench VerifiedSoftware Engineering Benchmark Verified | Score76% | Versus best verified row Best verified: Claude Opus 5 · 96% | Gap20 behind | WeightWeighted 16% | Provider exact |
| SciCodeScientific Code Benchmark | Score43.6% | Versus best verified row Best verified: Sakana Fugu · 60.1% | Gap16.5 behind | WeightWeighted 16% | Provider exact |
| SWE-bench Pro | Score51.2% | Versus best verified row Best verified: Claude Mythos 5 · 80.3% | Gap29.1 behind | WeightWeighted 10% | Provider exact |
| Terminal-Bench 2.1Terminal-Bench 2.1 (provider run) | Score51.7% | Versus best verified row Best verified: GLM-5.3 · 88.2% | Gap36.5 behind | WeightDisplay only | Provider exact |
Agentic4 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| OSWorld-Verified | Score65.9% | Versus best verified row Best verified: Qwen3.8 Max · 86.1% | Gap20.2 behind | WeightWeighted 34% | Provider exact |
| MCP Atlas | Score75.5% | Versus best verified row Best verified: Muse Spark 1.1 · 88.1% | Gap12.6 behind | WeightDisplay only | Provider exact |
| DeepSearchQA | Score74.6% | Versus best verified row Best verified: Claude Opus 5 · 95.0% | Gap20.4 behind | WeightDisplay only | Provider exact |
| skillsBench | Score44.3% | Versus best verified row Best verified: Qwen3.8 Max · 70.2% | Gap25.9 behind | WeightDisplay only | Provider exact |
Math1 row
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| AIME26AIME 2026 | Score94.7% | Versus best verified row Best verified: GLM-5.2 · 99.2% | Gap4.5 behind | WeightWeighted 25% | Provider exact |
Multimodal4 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| MMMU-ProMassive Multi-discipline Multimodal Understanding Pro | Score74% | Versus best verified row Best verified: GPT-5.4 Pro · 94% | Gap20 behind | WeightWeighted 45% | Provider exact |
| CharXivCharXiv Reasoning | Score78.8% | Versus best verified row Best verified: Claude Mythos 5 · 93.5% | Gap14.7 behind | WeightWeighted 25% | Provider exact |
| ScreenSpot Pro | Score75.4% | Versus best verified row Best verified: Claude Opus 4.8 · 87.9% | Gap12.5 behind | WeightDisplay only | Provider exact |
| OmniDocBench 1.5 | Score75.8% | Versus best verified row Best verified: Qwen3.8 Max · 92.1% | Gap16.3 behind | WeightDisplay only | Provider exact |
Inst. Following1 row
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| IFBenchInstruction Following Benchmark | Score77% | Versus best verified row Best verified: MAI-Thinking-1 · 85% | Gap8 behind | WeightWeighted 65% | Provider exact |
Lineage
The sequence follows explicit supersedes links. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
Aug 10, 2026 · you are here
Muse Glimmer 30BScore 52.4 · Price not listed
30b · 30B
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
Muse Glimmer 30B ranks #116 of 224 on the public leaderboard with a score of 52.38/100. It does not yet have enough sourced coverage for a verified position.
Muse Glimmer 30B is a open weight model with a 131K context window. It uses an explicit reasoning mode, which can improve complex problem solving while adding latency and token use.
Meta publishes the full-precision and two 4-bit Muse Glimmer 30B checkpoints under Apache 2.0 on Hugging Face. The K-Quant-17GB release targets 24 GB hardware, while K-Quant-Dynamic targets 32 GB; Meta also ships a DFlash speculative-decoding drafter and a dedicated perception encoder.
14 of 402 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Instruction Following at #21, while its lowest eligible position is Coding at #62. a well-rounded choice across a range of tasks.
Radar
Full release historyMuse Glimmer 30B release history
Muse Glimmer 30B released
Meta · Model release
Source confirmed
Frequently asked questions
How does Muse Glimmer 30B perform overall in AI benchmarks?
Muse Glimmer 30B ranks #116 out of 224 models on the public BenchAlign leaderboard, with a score of 52.38/100. Its evidence status is Estimated, and this profile shows 14 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is Muse Glimmer 30B good for coding and programming?
Muse Glimmer 30B ranks #62 out of 141 eligible models for coding and programming, with a public category score of 52/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Muse Glimmer 30B good for mathematics?
Muse Glimmer 30B has source-displayable benchmark coverage for mathematics, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Muse Glimmer 30B good for agentic tool use and computer tasks?
Muse Glimmer 30B ranks #53 out of 137 eligible models for agentic tool use and computer tasks, with a public category score of 51.7/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Muse Glimmer 30B good for multimodal and grounded tasks?
Muse Glimmer 30B ranks #27 out of 34 eligible models for multimodal and grounded tasks, with a public category score of 48.9/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Muse Glimmer 30B good for instruction following?
Muse Glimmer 30B ranks #21 out of 41 eligible models for instruction following, with a public category score of 84.5/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Muse Glimmer 30B open source?
Muse Glimmer 30B is an open-weight model from Meta. Its weights can be downloaded for local or hosted deployment, subject to the published license. Open weight does not automatically mean open source: training data and training code may remain private, and commercial restrictions can still apply.
Does Muse Glimmer 30B have full benchmark coverage on BenchLM?
No. Muse Glimmer 30B currently has 30 source-displayable rows across 402 tracked benchmark slots. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of Muse Glimmer 30B?
Muse Glimmer 30B has a reported context window of 131K in the exact-model catalog record. The value stays visible, but the profile marks its source link as unavailable instead of presenting it as directly documented. Maximum output length remains separate because providers often publish a different limit.
Last updated August 22, 2026. Runtime fields remain blank until a sourced snapshot exists.
Watch Muse Glimmer 30B in the weekly brief
Get one weekly email when material rank, price, availability, or benchmark evidence changes are worth revisiting.
Read a sample issueJoin 2,000+ readers.