Muse Glimmer 30B
Muse Glimmer 30B is Meta's August 2026 open-weight model from Meta Superintelligence Labs, released under Apache 2.0. It is a dense model of about 30B parameters (including a 1.8B vision encoder) built for agentic tool use. Meta targets 24-32 GB machines with 4-bit builds and a speculative-decoding drafter.
Specifications
- Context window
- 128Kabout 98,000 words, or 200 pages
- Max output
- Not published
- Input price
- $0.35per 1M tokens
- Output price
- $1.50per 1M tokens
- Cached input
- $0.04per 1M tokens
- Weights
- OpenApache-2.0
- Parameters
- 29.8Bdense
- Input types
- Text, Images, Video
- Output types
- Text
- Reasoning mode
- Yes
- Tool calling
- Yes
- Released
- Aug 10, 2026Current
Together AI serverless price (model meta-models/Muse-Glimmer-30B); the $0.04 cached-input rate is from the Together pricing page. Meta's Model API page does not list a Glimmer price.
What it costs in practice
| Workload | Monthly cost at list price |
|---|---|
| Support chatbot30,000 conversations a month, 2,000 input and 400 output tokens each | $39 |
| Document summaries2,000 documents a month, 12,000 input and 800 output tokens each | $11 |
| Coding agent1,000 tasks a month, 60,000 input and 6,000 output tokens each | $30 |
Standard rates, no caching or batch discounts. Run your own numbers.
Strengths
- Apache 2.0 license with no usage thresholds
- Adjustable reasoning effort (low to xhigh)
- Official 4-bit builds sized for 24-32 GB GPUs and Macs
Trade-offs to weigh
- 131K context is shorter than many 2026 open models
- BF16 weights exceed 55 GB, so most local setups need quantization
Where it fits
Meta positions Muse Glimmer 30B for:
Where you can use it
- Hugging Face
- Ollama
- Together AI
- Fireworks
- OpenRouter
- LM Studio
Run it yourself
Released under Apache-2.0. Approximate memory needed just to load the weights:
| Precision | Memory for weights |
|---|---|
| 16-bit (BF16/FP16) | 60 GB |
| 8-bit (FP8/INT8) | 30 GB |
| 4-bit (Q4_K_M, about 4.85 bits) | 18 GB |
Add memory for the KV cache (it grows with context length and concurrent users) and 1-2 GB of runtime overhead. Size it precisely in the VRAM calculator.
Alternatives to consider
| Model | Tier | Context | Input / output per 1M | |
|---|---|---|---|---|
| Gemini 3.1 Flash-LiteGoogle | Efficient | 1M | $0.25 / $1.50 | Compare |
| SonarPerplexity | Efficient | — | $0.25 / $2.50 | Compare |
| Gemini 3.5 Flash-LiteGoogle | Efficient | 1M | $0.30 / $2.50 | Compare |
| Codestral 25.08Mistral AI | Efficient | 125K | $0.30 / $0.90 | Compare |
Frequently asked questions
How much does Muse Glimmer 30B cost?
Muse Glimmer 30B lists at $0.35 per million input tokens and $1.50 per million output tokens, with cached input at $0.04 (checked October 10, 2026). Together AI serverless price (model meta-models/Muse-Glimmer-30B); the $0.04 cached-input rate is from the Together pricing page. Meta's Model API page does not list a Glimmer price. Confirm current rates on the provider's pricing page before you commit.
What is the context window of Muse Glimmer 30B?
Muse Glimmer 30B accepts up to 131,072 tokens of context (about 98,000 words, or 200 pages). Long prompts cost more and can be slower, so most apps send only what the task needs.
Is Muse Glimmer 30B open source?
Muse Glimmer 30B is an open-weight model released under the Apache-2.0 license. You can download the weights from Hugging Face (meta-models/Muse-Glimmer-30B) and run or fine-tune them yourself, within the license terms. Open weights isn't the same as fully open source: training data and code may not be published.
Can I run Muse Glimmer 30B locally?
Yes, if you have enough memory. The weights alone need about 18 GB at 4-bit quantization and about 60 GB at 16-bit, plus room for the KV cache. Use the VRAM calculator on LLModel.com to check a specific GPU or Mac.
What is Muse Glimmer 30B best for?
Meta positions Muse Glimmer 30B for agents, on-device, self-hosting, coding. It accepts text, images, video as input, has a reasoning (thinking) mode and supports tool calling. Test it on a sample of your own tasks before you choose it for production.
Sources
- dev.meta.ai/models/muse-glimmer
- docs.together.ai/docs/serverless-models
- research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model
- huggingface.co/meta-models/Muse-Glimmer-30B
- huggingface.co/meta-models/Muse-Glimmer-30B/raw/main/config.json
- huggingface.co/api/models/meta-models/Muse-Glimmer-30B
- www.together.ai/pricing
- ollama.com/library
Facts checked against these pages on October 10, 2026. Spot an error? Tell us.
Still choosing a model?
Send us your use case and constraints. We'll come back with a shortlist and an evaluation plan you can run on your own data.
- Free and vendor-neutral: no model provider pays for placement.
- Your details go only to partners you agree to.