Meta Muse

Meta Muse

Usage Based

Meta's model family for agents, code, and multimodal work, closed and open.

LLM
Proprietary

Published 1 October 2026Last updated 3 November 2026

Scores

Popularity4/5

Widely used

Meta AI puts Muse Spark in front of billions of consumers, and each release is covered widely in tech press. As a developer API it only opened in July 2026, so its footprint in production codebases is still small next to OpenAI, Claude, or Gemini.

Learning Curve2/5

Easy

The Meta Model API follows the OpenAI chat-completions format, so anyone who has called a hosted LLM can switch to it by changing the base URL and model ID. Choosing between reasoning levels and the Standard or Contributor terms is the only new decision.

Flexibility3/5

Configurable

Covers text, image, and video input, tool calling, and adjustable reasoning effort, but there are no weights to fine-tune or self-host and customization stops at what the API exposes.

Performance4/5

Fast

The 1.3 release posts large jumps on computer-use and knowledge-work evaluations (OSWorld 2.0 from 47.6 to 66.9) and near-perfect long-context retrieval out to 1M tokens, while Meta's own charts still place it just behind Claude Opus 5 on the hardest agentic tasks.

Portability2/5

Mostly locked in

Closed weights and a US-only first-party API tie workloads to Meta, though the OpenAI-compatible format and availability on OpenRouter keep switching costs low.

About Meta Muse

Muse is the model family built by Meta Superintelligence Labs (MSL), the division Meta formed after Llama 4's reception in 2025. Muse Spark launched on April 8, 2026 inside Meta AI, where it replaced Llama as the model behind meta.ai, the Meta AI app, and the assistant in WhatsApp, Instagram, and Messenger. The family now spans closed models served by Meta and open-weight models anyone can download, while Llama remains a separate, older open-weight line.

On the closed side, Muse Wonder is the most capable model, and Muse Spark is the faster, cheaper workhorse, with versions 1.1, 1.2, and 1.3 still callable. Both are natively multimodal reasoning models tuned for tool use, computer use, long-horizon coding, and knowledge work, with a 1,048,576-token context window and selectable reasoning effort. Developers reach them through the Meta Model API, an OpenAI-compatible endpoint with usage-based token pricing, and through routers such as OpenRouter; they also appear in the model pickers of coding agents including OpenCode and Cline, and power Meta's own Muse Code agent.

Spark 1.2 and 1.3 also come in a Contributor mode: the same models at a steep discount in exchange for Meta using the prompts and completions to train future models, with tighter rate limits. It is why Muse Spark shows up as a near-free option in coding tools, and the main privacy trade-off to check before routing production traffic to it.

The open-weight side has two models. Muse Spark Open is an 816B-parameter mixture-of-experts model with 42B active parameters and a 1M-token context, released under the Muse Model License and served by major clouds and inference platforms; Muse Glimmer is a 30B model distilled from Muse Spark for local and edge use, released under Apache 2.0.

Key Features

  • Muse Wonder as the top closed model, with Muse Spark as the faster, cheaper tier
  • 1,048,576-token context window and native text, image, and video input
  • Selectable reasoning effort, tuned for tool use, computer use, and agentic coding
  • OpenAI-compatible Meta Model API with prompt caching
  • Discounted Contributor mode on Spark 1.2 and 1.3 in exchange for training data
  • Muse Spark Open: 816B-parameter open-weight model with 42B active parameters
  • Muse Glimmer: 30B open-weight model for local and edge deployment
  • Powers Meta AI, the Muse personal agent, and the Muse Code coding agent

Pros

  • Muse Spark's standard token pricing sits well below the other US frontier labs for comparable agentic results
  • One family covers a hosted API and self-hostable open weights
  • OpenAI-compatible endpoint means existing SDKs and agent frameworks work with a base URL change
  • Contributor mode makes it free or near-free to try inside OpenCode and Cline

Cons

  • Muse Spark Open needs a multi-GPU server, and its Muse Model License is not an open-source licence
  • Muse Wonder and the newest Spark versions are closed weights, served only by Meta and its partners
  • Contributor pricing sends prompts and completions to Meta for training, and some tools default users to it
  • Short track record as a developer API compared with OpenAI, Anthropic, and Google, with frequent version turnover

Meta Muse Pricing

Usage Based
Muse Spark Open and Muse Glimmer (self-hosted)Free
  • Free open weights on Hugging Face
  • Muse Spark Open under the Muse Model License; Muse Glimmer under Apache 2.0
  • You pay only for your own GPUs or a hosting provider
Meta AI (consumer)Free
  • Free chat use through meta.ai, the Meta AI app, WhatsApp, Instagram, and Messenger
  • No API access
Muse WonderContact sales
  • $4 per million input tokens, $20 per million output tokens
  • $0.20 per million cached input tokens
  • Top closed Muse model; no Contributor mode
Muse Spark StandardContact sales
  • $1.25 per million input tokens, $4.25 per million output tokens
  • $0.15 per million cached input tokens
  • Same rate for versions 1.1, 1.2, and 1.3
Muse Spark ContributorContact sales
  • $0.10 per million input tokens, $0.20 per million output tokens; cached $0.002
  • Versions 1.2 and 1.3 only; prompts and completions may train future Meta models
  • Lower rate limits than the Standard tier

Tools Related to Meta Muse

Works well with Meta Muse(1)

Alternatives to Meta Muse(5)

Muse Spark offers a 1M-token context window, video input, and a discounted Contributor tier through Meta's OpenAI-compatible API; Claude has the longer track record with developers and far wider availability, since Meta's own API is still a US-only preview.

Muse Spark's Meta Model API follows the OpenAI chat-completions format, so trying it is a base-URL change. OpenAI has the more mature developer platform and worldwide availability; Muse Spark adds video input and a Contributor tier that trades prompts for a lower price.

Muse Spark and Grok are the newer challengers among proprietary models, each tied to a social platform: Muse Spark powers Meta AI across WhatsApp and Instagram, while Grok draws on real-time X data. Both expose OpenAI-compatible APIs.

Muse Spark and Google Gemini are both natively multimodal models with 1M-token context windows and video input. Gemini is available worldwide through Google AI Studio and Vertex AI, while Meta's first-party API is a US-only preview reached elsewhere through routers.

Meta's two model lines: Muse Spark is the closed-weight frontier family reached only through an API, while Llama stays open-weight and can be downloaded, fine-tuned, and self-hosted. Pick Llama for control over the weights, Muse Spark for Meta's strongest model.

Vendor

Tags

Web

Details

Maintained
Yes