meta-models/Muse-Glimmer-30B · Hugging Face

Source: https://huggingface.co/meta-models/Muse-Glimmer-30B

https://huggingface.co/meta-models/Muse-Glimmer-30B#muse-glimmer-model-cardMuse Glimmer Model Card

**Authors:**Meta Superintelligence Lab **Model Release Date:**August 2026 License:Apache 2.0

Muse Glimmer is a 30-billion-parameter causal language model with a dedicated perception encoder, distilled from Muse Spark and purpose-built for autonomous agentic tasks on consumer hardware. The model integrates multi-step reasoning, reliable tool use, multimodal understanding, and failure recovery into a single model that runs locally without requiring cloud infrastructure or network access.

Building effective agents requires key capabilities working together to achieve the user’s goals. Muse Glimmer is trained and evaluated on these capabilities:

• **End-to-end Agentic Task Completion.**Muse Glimmer achieves strong success rates on full-task benchmarks including DeepSearch QA, MCP-Atlas, 𝛕3-Bench and SWE-Bench, which measure its ability to work within scaffolds, write and debug code, and resolve multi-turn requests from start to finish.
• **Reliable Tool Use.**The model handles a wide range of function calls, invoking tools with precise schemas throughout extended workflows.
• **Multi-Step Reasoning.**Muse Glimmer chains reasoning over long horizons, sustaining coherent plans across complex, extended workflows.
• **Failure Recovery.**When a tool call fails or returns an unexpected result, the model diagnoses the error and retries rather than halt.
• **Multimodal Input and Reasoning.**Through a dedicated perception encoder, the model accepts interleaved text and images. This enables agents to interpret screenshots, charts, and documents alongside conversation.
• **Scaffold Compatibility.**Muse Glimmer works across OpenClaw, Hermes Agent, and other agentic orchestration patterns.
• **Controllable Effort.**The model supports different reasoning strengths to select the right balance between quality and speed.
• **Multilingual.**Muse Glimmer is trained on data from more than 100 languages.

https://huggingface.co/meta-models/Muse-Glimmer-30B#muse-glimmer-30b-model-overviewMuse Glimmer-30B Model Overview

Model ArchitectureDense Causal Transformer with Perception EncoderTotal Parameters~29.6BLanguage Model****ArchitectureDense Causal TransformerNumber of Parameters29.6B (including vision encoder)Hidden dimension6656Layers52Attention pattern[Local, Local, Local, Global] repeatingSliding window size2048Gated attentionYesAttention heads (Q / KV)32 / 2 (GQA ratio 16:1)Head dimension128FFN typeSwiGLUFFN intermediate dimension19,968Position encodingRoPE (θ = 500,000), local layers onlyPerceptionencoder~1.8B param ViT-G/14, 50 layers, width 1536, patch size 14Vocabulary size202,048Tokenizer200,000 BPE tokens + 2,048 special tokensMax visual tokens per image4,096Context length131,072+Supported modalitiesInput: text + image, Output: textTraining DataMultimodal content sourced from publicly available data, data provided by third parties and information from Meta’s products and services, curated and enriched by external vendor networks and Meta personnel.Knowledge cutoffJanuary 4, 2026

https://huggingface.co/meta-models/Muse-Glimmer-30B#optimized-for-local-deploymentsOptimized for Local Deployments

Muse Glimmer was optimized for local deployment, and designed to run at practical speeds on consumer hardware without sacrificing quality.

**Fitting the Model on Your Device.**We use quantization techniques to compress the model’s weights to approximately 4-bit precision, shrinking the language model to under 20 GB. This leaves enough headroom for the model’s KV cache, the perception encoder for image understanding, and the speculative decoding drafter to run simultaneously within a 24 GB or 32 GB envelope. Critically, we validated that this compression introduces minimum to no degradation on agentic tasks.

Full PrecisionK-Quant-DynamicK-Quant-17GB% Degradation*-0.2%1.0%Target Hardware64GB VRAM32GB VRAM24GB VRAM * Degradation measured using an average on accuracy metrics across 15 common benchmarks

Faster Generation Through Speculative DecodingMuse Glimmer ships with a lightweight “drafter” model based onDFlash, a small companion network that proposes entire blocks of tokens at once. The DFlash block-diffusion model predicts entire blocks of 16 tokens in a single forward pass. The main model then verifies these proposals in parallel, accepting correct tokens and correcting wrong ones. This technique lets Muse Glimmer generate text significantly faster than standard token-by-token generation while producing identical output quality. We provide quantized drafter versions to incur a smaller memory overhead in the release.

ComponentSettingDraft layers5Block size16AttentionSliding-window, 2048, all layersAttention heads32 query / 8 KV (GQA)Sequence length131,072Hidden-feature layers5, uniform over target: {1, 13, 25, 37, 49} of 52 We measure the speed of our K-Quant-17GB model alongside the quantized DFlash drafter on MacBook M4-Max, M5-Max and on an Nvidia RTX-5090. The model is fast enough for fluid conversation and real-time agent interaction, all running entirely on your device.

GPUBaseline No-speculation (tok/s)Avg* with DFlash Speculation (tok/s)SpeedupNvidia RTX 509074.9233.43.1xApple M4 Max23.737.81.5xApple M5 Max26.650.21.8x * Average across a diverse prompt set. Measurements done with batch size 1 and greedy decoding. M4/M5 measurements were done using ExecuTorch, and RTX using llama.cpp.

https://huggingface.co/meta-models/Muse-Glimmer-30B#benchmarksBenchmarks

We evaluated Muse Glimmer across a broad range of benchmarks to assess the diverse capabilities required for effective autonomous agent behavior. Compared with Gemma4-31B and Qwen3.6-27B, Muse Glimmer performs strongly for its size class on several widely used LLM benchmarks.

CategoryBenchmarkMuse Glimmer-30B High ReasoningGemma4-31B Thinking ModeQwen3.6-27B Thinking ModeGeneral AgenticMCP Atlas (Public)75.554.262.5DeepSearch QA74.661.771.1𝛕3-Banking23.515.116.7WildClawBench47.637.643.2GDPVal-AA v29538111141Gaia243.336.440.0SkillsBench (with skills)44.332.446.6OSWorld-Verified65.958.575.6Agentic CodingSWE-Bench Pro51.236.950.2SWE-Bench Verified76.066.677.2TerminalBench 2.1 (with terminus2)51.743.460.7SciCode43.643.439.8MultimodalCharxiv Reasoning78.877.778.4ScreenSpot Pro75.475.976.1OmniDocBench v1.575.872.577.8MMMU Pro747375SafetyCI MemoriesViolation (↓): 26.4 Coverage: 64.8Violation (↓):12.1 Coverage: 53.0Violation (↓): 53.4 Coverage:66.9Siren AgentDojoAttack Success Rate (↓): 28.4 Utility:94.2Attack Success Rate (↓):25.6 Utility: 90.8Attack Success Rate (↓): 40.3 Utility: 92.7General Capabilities and ReasoningIFBench77.076.070.8AIME 202694.789.294.1GPQA Diamond (AA)83.585.784.2HLE Text (AA)22.023.623.1AA-LCR80.068.373.3Beam128K65.158.263.0 For more detail about our evaluations, see ourreport.

https://huggingface.co/meta-models/Muse-Glimmer-30B#best-practicesBest Practices

To achieve best performance, we recommend the following settings:

**Sampling Parameters:**Use the following configuration:

• temperature = 1.0
• top_p = 0.95
• top_k = 64

**Reasoning Strength:**Reasoning strength controls how much the model thinks before responding to the prompt. Reasoning strength can be defined as part of the system prompt asReasoning strength: <value\>. Muse Glimmer supports the following levels: low / medium / high / xhigh. Use high or xhigh for complex problem solving, coding, and agentic tasks.

https://huggingface.co/meta-models/Muse-Glimmer-30B#trust-and-safetyTrust and Safety

As we would for other large language models, we strongly recommend that Muse Glimmer be deployed not as an endpoint in itself but as part of an overall AI system with additional guardrails as required or appropriate for the use cases and context of its deployment. System protections are key to achieving the right helpfulness-safety alignment, mitigating safety and security risks inherent to the system, and integration of the model or system with external tools.

Evaluations We evaluated Muse Glimmer for common use cases as well as specific capabilities. Common use cases evaluations measure safety risks of systems for most commonly built applications including chat bot and visual, QA. We built dedicated, adversarial evaluation datasets and evaluated systems composed of Muse Glimmer models and those safeguards to filter input prompt and output response. It is important to evaluate applications in context, and we recommend building dedicated evaluation datasets for your use case.

Capability evaluations measure vulnerabilities of models inherent to specific capabilities, for which were crafted dedicated benchmarks. We also used industry standard safety and capability benchmarks where appropriate.

Muse Glimmer was primarily evaluated across four risk axes:

1. Content safety— Standard alignment for refusal of harmful requests and calibrated responses to borderline prompts.
2. Agentic risk— Policies for irreversible-action confirmation, data minimization, scaffold boundary respect, and indirect prompt-injection resistance.
3. Privacy (Appropriate Information Flows)— Respect for contextual integrity of information when interacting with third parties on an individual’s behalf, inspired by CI theory.
4. Preparedness— Chemical & biological, cyber, and loss-of-control risks.

https://huggingface.co/meta-models/Muse-Glimmer-30B#preparednessPreparedness

Muse Glimmer does not fall under the definition of “Frontier AI” in Meta’s Advanced AI Scaling Framework (AAISF), since it is generally less capable than Muse Spark. However, as a matter of prudence, our Preparedness Team assessed Muse Glimmer’s risk profile and determined that it would receive the following designations:

• Chem/Bio: Moderate or lower risk;
• Cyber: Moderate or lower risk (inferred);
• Loss of Control: Moderate or lower risk (inferred).

Cyber and Loss of Control risk levels are inferred to be Moderate or lower since Muse Glimmer is broadly weaker than Muse Spark 1.0, which received the same risk designation in these domains.

In the chem/bio domain, we evaluated Muse Glimmer on a range of benchmarks for scientific knowledge and wet-lab debugging (most performant in Muse Glimmer’s size class are bolded; second most performant is underlined — Kimi K3 is also included for context):

BenchmarkMuse Glimmer-30BGemma4-31BQwen3.6-27BKimi K3MBCT41.5%**50.6%**45.9%*58.9%*HPCT52.3%**54.0%**48.7%*59.6%*VCT37.0%**43.5%**33.7%*48.0%*WMDP (Bio)**86.5%**85.9%84.8%*89.1%*WMDP (Chem)75.2%**80.5%**74.8%*84.2%*Lab Bench (ProtocolQA)**80.2%**75.8%69.1%81.9% We find that Muse Glimmer’s abilities are approximately in line with other models in its size class, while showing strictly lower capabilities than larger open-weight models, suggesting that it is unlikely to materially enable new threats upon release. We also evaluated it on our suite that focuses on the unique set of bottlenecks that would otherwise deter or limit the success of real-world threat actors; here, our evaluation rated its risk rating at moderate or lower as well. See theMuse Spark Safety & Preparedness Reportfor a detailed description of the above evaluations and our methodology.

https://huggingface.co/meta-models/Muse-Glimmer-30B#train-time-mitigationsTrain-Time Mitigations

1. **Safety SFT:**Curated examples demonstrating correct safety behavior, including agentic safety scenarios covering tool-use boundaries, prompt injection resistance, and permission handling.
2. **Safety RL:**Reinforcement learning with safety-specific reward signals that penalize policy violations while rewarding helpful responses to legitimate requests.
3. **Appropriate information flows:**Principles of data sensitivity recognition, minimization, and local-first execution embedded directly into model weights through dedicated synthetic training data.

https://huggingface.co/meta-models/Muse-Glimmer-30B#intended-useIntended Use

**Intended Use Cases:**Muse Glimmer is intended for commercial and research use. The model is optimized for autonomous agentic tasks including:

• **Local AI agents:**Multi-step planning, sequential tool invocation, failure recovery, and long-horizon task execution running entirely on consumer devices.
• **Coding agents:**Writing, debugging, and resolving real-world software engineering tasks (e.g., SWE-Bench style workflows).
• **Tool use and function calling:**Reliable schema-based tool invocation across extended, multi-turn workflows.
• **Multimodal reasoning:**Interpreting screenshots, charts, documents, and images alongside conversation for agentic and information-rich environments.
• **Synthetic data generation:**Generating high-quality training data for downstream model development.
• **LLM-as-a-judge evaluation:**Serving as an evaluator for other models’ outputs.

**Out-of-scope:**Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Apache 2.0 License terms. Audio input/output is not supported.

https://huggingface.co/meta-models/Muse-Glimmer-30B#considerations-and-limitationsConsiderations and Limitations

Muse Glimmer is a technology that carries known and unknown risks. Testing conducted to date has not, and could not, cover all scenarios.

Limitations:

• The model may produce inaccurate, biased, or objectionable responses to user prompts.
• While optimized for agentic tasks, the model may still make errors in multi-step reasoning, particularly in novel scenarios not well represented in training data.
• The model is not explicitly optimized for video; video input is processed as individual frames.
• The model has not been evaluated on all languages contained in the pre-training data. Performance may degrade on languages outside the strongly supported set.
• Quantized inference may show minor quality differences in edge cases compared to full-precision.
• The model is not intended to be downloaded by or used by individuals under the age of 18. Where deployed within systems that may be used by individuals under the age of 18, deployers are responsible for ensuring that any risks associated with such use by individuals under the age of 18 has been fully assessed and appropriately mitigated, and complies with all applicable laws.

**Responsible Use:**Developers should perform their own safety testing and tuning tailored to their specific applications and proposed languages. Our Usage Policy can be found here [link]. We recommend implementing additional guardrails (such as human-in-the-loop confirmation for irreversible actions) when deploying the model in agentic contexts where it can take real-world actions.

https://huggingface.co/meta-models/Muse-Glimmer-30B#released-artifactsReleased Artifacts

All artifacts are released under Apache 2.0:

ArtifactDescriptionFull-precision weights (BF16)Complete model weights for fine-tuning and research4-bit quantized weights (2 variants)Optimized for inference on 24/32 GB consumer hardwareDFlash drafter headSpeculative decoding companion for faster generationPerception encoderFrozen ViT-G/14 vision encoder (~1.8B params) **Where to send questions or comments about the model:**Please provide any feedback, comments or bug reports on the model through the Hugging Face page athttps://huggingface.co/meta-models/. For more technical information about generation parameters and recipes for how to use Muse Glimmer in applications, please see the developer documentation.