Collect Video Testimonials That Build Trust
Muse Wonder
Meta Muse Wonder is a multimodal AI model for autonomous agents, complex reasoning, software engineering and computer use, with a 1 million token context window and API access.
About Muse Wonder
Muse Wonder is an artificial intelligence model developed by Meta for complex software engineering, reasoning and personal and professional AI agents. Officially released on October 12, 2026, it is designed to perform tasks that involve planning, tool interaction, multiple execution steps and sustained work toward specific goals. Its API model identifier is muse-wonder-1.0.
The model supports text, image, video, audio and PDF inputs, with text as its output modality. It provides a context window of 1,048,576 tokens and configurable reasoning effort from minimal to maximum. Supported tasks include document analysis, content generation, complex problem solving, coding, debugging, research assistance and multimodal understanding.
Muse Wonder incorporates Active Memory and Active Focus capabilities. Active Memory helps preserve and retrieve relevant information across conversations, while Active Focus supports sustained execution of challenging tasks. The model can plan workflows, delegate tasks to subagents, invoke tools in parallel and recover from execution errors. It also supports computer use through compatible agent environments.
Developers can access Muse Wonder through Meta Model API and Muse Code. The model supports function calling, structured outputs, search grounding and integration with compatible development tools. It is intended for commercial and research applications, including software engineering, business automation, technical research and AI agent development.
Pricing
-
Standard API Input Tokens: $4.00 per 1 million input tokens Applies to Muse Wonder requests through Meta Model API in Standard mode.
-
Standard API Output Tokens: $20.00 per 1 million output tokens Applies to tokens generated by Muse Wonder through Meta Model API.
-
Cached Input Tokens: $0.20 per 1 million cached input tokens Applies when eligible prompt content is served from the prompt cache.
-
Web Search Grounding: $2.50 per 1,000 search queries Additional usage charge when invoking the supported web search tool. Token charges also apply.
-
Muse Code Power Usage: $50 per month Includes access to Muse Wonder Full through Muse Code, with 20 times the usage allowance of the Everyday Usage plan.
-
Muse Code Maximum Usage: $200 per month Includes access to Muse Wonder Full through Muse Code, with 80 times the usage allowance of the Everyday Usage plan.
Muse Wonder is available through Meta Model API using pay as you go billing in Standard mode. Contributor mode is not available for Muse Wonder through the pay as you go API.
Muse Wonder Full is a separate model variant available through eligible Muse Code subscriptions and is not available through the pay as you go Model API.
Use Cases
- Building autonomous AI agents that plan, execute and monitor complex workflows involving external tools and applications.
- Developing software engineering agents for code generation, debugging, testing, repository maintenance and multi file modifications.
- Creating computer use agents that interact with browsers, desktop applications and other supported digital interfaces.
- Analyzing documents, images, video and large information sources for research, professional knowledge work and complex decision support.
Pros & Cons
Pros:
- Supports a 1,048,576 token context window and multimodal inputs, including text, images, video, audio and PDF documents.
- Provides agentic capabilities including task planning, subagent delegation, parallel tool calling, sustained execution and error recovery.
- Supports software engineering, configurable reasoning, structured outputs and API integration through compatible development interfaces.
Cons:
- Produces text output only. Native image, video and audio generation are not supported by Muse Wonder.
- Visual and spatial reasoning, especially tasks requiring precise three dimensional understanding, remains a documented limitation.
- May produce inaccurate responses or make errors during complex multistep reasoning. API usage is subject to per team limits of 3,000 requests per minute and 4 million tokens per minute.
Integrations
Meta Model API, Muse Code, OpenRouter, Responses API, Chat Completions API, Messages API, OpenAI SDK compatible clients, Anthropic SDK compatible clients, OpenAI compatible agent CLIs, Function Calling, Structured Outputs, Web Search Grounding, Computer Use, OpenCode
FAQ
What is Muse Wonder?
Muse Wonder is an AI model developed by Meta for complex reasoning, software engineering and personal and professional AI agents. It supports long running tasks, tool calling, multimodal understanding and multistep workflows.
When was Muse Wonder released?
Meta officially released Muse Wonder on October 12, 2026. The documented API model identifier is muse-wonder-1.0.
What is the context window of Muse Wonder?
Muse Wonder supports a context window of 1,048,576 tokens. This allows developers to provide extensive documents, code repositories and conversation context within a single model request.
Does Muse Wonder support images, video and audio?
Yes. Muse Wonder accepts text, images, video, audio and PDF documents as input and produces text as output. It supports multimodal understanding but is not a native image, video or audio generation model.
Can Muse Wonder be used to build autonomous AI agents?
Yes. Muse Wonder is designed for agentic tasks involving planning, tool execution, subagent delegation, error recovery and sustained workflows. Developers can connect the model to external tools and applications through supported interfaces and agent environments.
Does Muse Wonder support coding and software development?
Yes. Muse Wonder supports code generation, debugging, testing, explanation and modification. It can work across multiple files in software repositories and assist with extended software engineering tasks involving implementation and verification.
Does Muse Wonder support function calling and structured outputs?
Yes. Muse Wonder supports tool calling and structured outputs through Meta Model API. Developers can connect external functions to the model and request outputs that conform to supported structured formats.
Can Muse Wonder interact with browsers and desktop applications?
Yes. Muse Wonder supports computer use through compatible tools and agent environments. It can support agents that interact with browsers and desktop applications. Actual actions depend on the available tools, permissions and execution environment.
What is the difference between Muse Wonder and Muse Wonder Full?
Muse Wonder and Muse Wonder Full are separate model variants. Muse Wonder is accessible through Meta Model API and Muse Code. Muse Wonder Full is a higher capability variant available through eligible Muse Code subscriptions. Selecting maximum reasoning effort on Muse Wonder does not switch the request to Muse Wonder Full.
What are the limitations of Muse Wonder?
Muse Wonder can produce inaccurate or biased responses and may make mistakes during complex multistep reasoning. Meta identifies precise visual and spatial reasoning, particularly three dimensional tasks, as a limitation. Performance may vary across languages and deployment configurations. The model produces text output and is not intended for individuals under 18.
Tags:
Last edited
October 13, 2026 at 5:33 AM by Venkatraman Chandrasekaran
Similar to Muse Wonder
