# Muse Spark 1.1

Muse Spark 1.1 plans and orchestrates work across tools and services. It reads text, images, video, PDFs, and audio, holds a context window of 1.0M tokens, and runs as a main agent or as a subagent.

- **Model ID:** `meta/muse-spark-1.1`
- **Type:** chat
- **Providers:** meta
- **Context window:** 1,048,576
- **Maximum output tokens:** 1,048,576
- **Pricing:** $1.25/1M input tokens, $4.25/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/muse-spark-1.1

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { streamText } from 'ai'

const result = streamText({
  model: 'meta/muse-spark-1.1',
  prompt: 'Why is the sky blue?'
})
```

## About

Muse Spark 1.1, released July 9, 2026, is a multimodal reasoning model from Meta built for agentic tasks. It accepts text, image, video, PDF, and audio input, and supports a context window of 1.0M tokens.

Muse Spark 1.1 plans work and then routes it across the tools and services you expose. It holds either position in an agent hierarchy: a main agent that delegates, or a subagent that takes an assignment and reports back. That flexibility matters when you compose a system from several models and want one of them coordinating the rest.

Muse Spark 1.1 works with new tools, Model Context Protocol (MCP) servers, and custom skills without examples. You describe the interface and Muse Spark 1.1 uses it, so adding a tool doesn't mean writing few-shot demonstrations for every call pattern. Parallel tool calling runs independent calls at once instead of in sequence, which shortens the critical path in a multi-tool step.

Structured output keeps responses machine-readable, so a downstream stage parses a schema instead of scraping prose. Built-in search returns citations with its results, which gives you a source to check when an answer depends on current information.

Mixed input types cut preprocessing. A product spec that arrives as a PDF, a bug report with a screen recording attached, and a voice note from a customer call all go into the same request. None of them need a separate extraction step before Muse Spark 1.1 sees them.

Set the model to `meta/muse-spark-1.1` and call it with the AI SDK or any supported interface like Chat Completions, Responses, or Messages. AI Gateway routes across Meta with automatic failover. Muse Spark 1.1 supports a context window of 1.0M tokens and completions up to 1.0M tokens per request, at $1.25 per million input tokens and $4.25 per million output tokens.

## What to consider

Muse Spark 1.1 accepts five input types, and each one consumes context differently, so a request that mixes video and PDFs reaches the window of 1.0M tokens faster than a text prompt of similar apparent length. Budget tokens against representative payloads rather than word counts. Parallel tool calling means several calls can land at once, so confirm your tool handlers tolerate concurrent invocation. Built-in search attaches citations to its results, so decide early how your product surfaces those sources to users.

## When to use

### Best For

- Agentic systems where one model plans work and delegates it across tools and services
- Subagent roles inside a larger orchestration that hands off tasks and collects results
- Pipelines that add new tools, MCP servers, or custom skills without writing examples first
- Mixed-media requests where PDFs, video, audio, and screenshots arrive together
- Research steps that need citations attached to search results

### Consider Alternatives When

- The workload is text-only and a Llama model covers it at lower cost
- You need a context window beyond 1.0M tokens. Llama 4 Scout supports 10M tokens
- The task is a single-turn prompt with no tools, where orchestration adds overhead
- Cost per token dominates and a smaller Llama model meets your quality bar

## Best for

- **Agentic Orchestration:** Planning work and delegating it across the tools you expose
- **Subagent Roles:** Handling delegated assignments inside a larger multi-agent system
- **Zero-Example Tool Adoption:** New tools, MCP servers, and custom skills without few-shot demonstrations
- **Mixed-Media Requests:** PDFs, video, audio, and images arriving in a single call
- **Cited Research Steps:** Built-in search that returns sources alongside its results

## Consider alternatives

- **Text-Only Workloads:** A Llama model covers pure text pipelines at lower cost
- **Larger Context Needs:** Llama 4 Scout supports a 10M token window for bigger corpora
- **Single-Turn Prompts:** Orchestration and reasoning add overhead when no tools are involved
- **Cost-Sensitive Traffic:** A smaller Llama model may meet your quality bar for less

## Frequently asked questions

### What input types does Muse Spark 1.1 accept?

Text, image, video, PDF, and audio. All five go into the same request, so a spec document, a screen recording, and a voice note can arrive together without separate extraction steps.

### What makes Muse Spark 1.1 suited to agentic tasks?

Muse Spark 1.1 plans and orchestrates work across tools and services rather than answering in one shot. It runs as a main agent that delegates or as a subagent that reports back, and it supports parallel tool calling for independent steps.

### Does Muse Spark 1.1 need examples to use a new tool?

No. Muse Spark 1.1 works with new tools, Model Context Protocol (MCP) servers, and custom skills without examples. Describe the interface and Muse Spark 1.1 calls it.

### Does Muse Spark 1.1 support structured output?

Yes. Structured output constrains a response to a schema your code parses directly, which keeps downstream stages from scraping prose.

### How does built-in search work with Muse Spark 1.1?

Built-in search returns citations alongside its results, so you can check the source behind an answer. Use it for steps that depend on current information.

### What is the context window for Muse Spark 1.1?

1.0M tokens, with completions up to 1.0M tokens per request.

### How do I use Muse Spark 1.1 on AI Gateway?

Use the identifier `meta/muse-spark-1.1` with the AI SDK or any supported interface like Chat Completions, Responses, or Messages. AI Gateway routes across `meta` and handles failover automatically.

### Does Muse Spark 1.1 support zero data retention?

Yes, Zero Data Retention is available for this model. Zero Data Retention is offered on a per-provider basis. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

## Links

- [Model page](https://vercel.com/ai-gateway/models/muse-spark-1.1)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://developer.meta.com/ai/models/muse-spark/)


If you encounter incorrect information, can’t find what you need, need a capability Vercel doesn’t currently offer, or have any other feedback to report, please do it by following the instructions at https://vercel.com/feedback.md.