Meta · Open-weight
Checked as of 2026-10-23
Llama & Muse
Meta's open models. Llama 4 is the previous generation; the new line is called Muse and lives in a different HF org.
At a glance
- Licence
- Open, restricted Muse Model License / Llama 4 Community License / Apache-2.0
Varies by model. Muse Glimmer is Apache-2.0. Muse Spark Open must be credited 'Built with Muse Spark Open', and model-as-a-service operators above US$20M revenue need a separate agreement. Llama 4 needs a licence from Meta above 700M monthly users, plus attribution and an acceptable-use policy.
- Access
- HFBedrockSageMaker JumpStart
- Hugging Face
- huggingface.co/meta-models
Variants 4
| Model | Params | Context | Hardware tier | Gated | Inputs | Status |
|---|---|---|---|---|---|---|
Muse Spark Open
meta-models/Muse-Spark-Open-816B-A42B
Released 2026-10-15. Muse Model License. Context is the card's '1M at inference'. |
816B42B active | 1M | Multi-GPU ml.p5.48xlarge and up | Yes | text, image, video, audio | current |
Muse Glimmer 30B
meta-models/Muse-Glimmer-30B
Dense, Apache-2.0. Parameter count includes the vision encoder. |
29.6B | 131.072K | 80GB GPU ml.p4de.24xlarge / ml.p5.48xlarge | No | text, image | current |
Llama 4 Maverick
meta-llama/Llama-4-Maverick-17B-128E-Instruct
April 2025. Llama 4 Community License. On Bedrock and JumpStart. |
400B17B active | 1M | Multi-GPU ml.p5.48xlarge and up | Yes | text, image | legacy |
Llama 4 Scout
meta-llama/Llama-4-Scout-17B-16E-Instruct
April 2025. Llama 4 Community License. On Bedrock and JumpStart. |
109B17B active | 10M | Multi-GPU ml.p5.48xlarge and up | Yes | text, image | legacy |
"—" means the figure could not be verified from a primary source on 2026-10-23 and has been left blank rather than guessed. Hardware tier is from total parameters (all experts for mixture-of-experts models), since every weight must sit in memory. SageMaker rates on the open-weight page.
Strengths
- Muse Spark Open is frontier-scale open weights: a mixture-of-experts model taking text, image, video and audio, with a 1M-token context.
- Muse Glimmer 30B is dense, Apache-2.0 and ungated: the easy one to start with on a single GPU.
- Llama 4 Scout and Maverick remain on Bedrock and JumpStart, with a 10M and 1M context respectively.
- A large community of quantised builds (GGUF, FP8) appears within days of each release.
Watch out for
- Meta's open models now live under the meta-models org. Nothing newer than Llama 4 has appeared in meta-llama.
- Muse Spark Open and every Llama repo are gated: Meta approves access by hand.
- No Muse model is on Bedrock yet. Only Llama 4 and older are.