# Muse Spark Open

Muse Spark Open is Meta's open-weight multimodal reasoning model, a sparse mixture-of-experts with 816 billion total and 42 billion active parameters. It is the downloadable sibling of the hosted Muse Spark line, aimed at coding and long-horizon agentic work that you can run on your own hardware or any inference host. Its published context window is 1,048,576 tokens. The advertised input rate is $1.25 per million input tokens. The advertised output rate is $4.25 per million output tokens.

- Model identifier: ` meta/muse-spark-open `
- Creator: Meta
- Providers: Fireworks
- Model type: Text generation
- Canonical model page: [Muse Spark Open](<https://impossibl.com/meta/muse-spark-open>)
- API availability: Listed as serving through Impossibl.
- Creator release date: 2026-10-15

## Providers

Fireworks hosts Muse Spark Open. Impossibl routes your request intelligently and handles failover automatically when a provider is down.

| Provider | Endpoint location | Routes | Verified context window | Tool calling |
| --- | --- | --- | --- | --- |
| Fireworks | Not specified | 1 | Not verified | Not verified |

The model price applies whichever provider serves the request. Endpoint locations do not guarantee inference residency.

## Specs and Capabilities

- Context window: 1,048,576 tokens
- Maximum output tokens: Not published in the public catalog; this is separate from the context window.
- Inputs: text, image
- Outputs: text
- Tool calling: Not published

## API access

Use ` meta/muse-spark-open ` as the model identifier through the Impossibl API.

Published API endpoints:

- ` /v1/chat/completions `

[API quickstart](<https://impossibl.com/docs/agent-quickstart>) · [Models and billing documentation](<https://impossibl.com/docs/models>)

## Published API prices

All rates are in USD. An unpublished rate is unknown, not zero. Workspace-specific pricing, taxes, balance-purchase fees, billing adjustments, and tool charges can affect the amount paid.

| Usage | Published base rate | Billing unit |
| --- | --- | --- |
| Input | $1.25 | per million input tokens |
| Output | $4.25 | per million output tokens |
| Cache read | $0.15 | per million cached input tokens |
| Cache write (default retention) | Not published | per million tokens |
| Cache write (one-hour retention) | Not published | per million tokens |
| Audio input | Not published | per million audio input tokens |

## Benchmarks

Last checked: 2026-10-19T03:22:58.358Z.
Configurations are evaluated separately. API scores retain their original units. Unmeasured scores are omitted.

### Muse Spark Open (Xhigh)

Source: [Evaluation source](<https://artificialanalysis.ai/models/muse-spark-open>)

| Benchmark | Score |
| --- | --- |
| Intelligence Index | 38.87 |
| Omniscience Index | 24.65 |
| GPQA | 88.99% |
| Humanity’s Last Exam | 43.88% |
| SciCode | 58.45% |
| CritPt | 14.86% |
| AA-LCR | 78.33% |
| GDPval Elo | 1401.12 |
| GDP.pdf All-pass | 18.60% |
| AutomationBench Partial | 49.71% |
| MMMU-Pro | 80.92% |
| legal index (API) | 44.5 |
| economics index (API) | 51 |
| engineering index (API) | 40.4 |
| intelligence index (API) | 38.9 |
| strategy and ops index (API) | 42.7 |
| finance and accounting index (API) | 42.5 |

Full benchmark data and breakdowns: [Complete benchmark snapshot](<https://api.impossibl.com/v1/models/meta/muse-spark-open/benchmarks>).


## Availability and retirement

The model is listed as serving. No retirement is announced in the current public catalog.

## Frequently asked questions

### What is Muse Spark Open?

Muse Spark Open is Meta's open-weight multimodal reasoning model, a sparse mixture-of-experts with 816 billion total and 42 billion active parameters. It is the downloadable sibling of the hosted Muse Spark line, aimed at coding and long-horizon agentic work that you can run on your own hardware or any inference host.

Sources: [Meta AI Research: Introducing Muse Spark Open](<https://research.meta.ai/blog/muse-spark-open>); [OpenRouter: Muse Spark Open](<https://openrouter.ai/meta/muse-spark-open>); [Fireworks AI model API: muse-spark-open](<https://api.fireworks.ai/v1/accounts/fireworks/models/muse-spark-open>)

### How much does Muse Spark Open cost?

Muse Spark Open costs $1.25 per million input tokens and $4.25 per million output tokens through the Impossibl API. Cache reads cost $0.15 per million tokens. All rates are in USD.

### What is the context length of Muse Spark Open?

Muse Spark Open has a 1,048,576-token context window through the Impossibl API.

### What inputs and outputs does Muse Spark Open support?

Muse Spark Open accepts text and image as input. Muse Spark Open returns text.

### How do I use Muse Spark Open through an API?

Muse Spark Open is available through the Impossibl API. Use meta/muse-spark-open as the model identifier with /v1/chat/completions. The quickstart explains how to create an API key and send your first request.

### When was Muse Spark Open released?

Meta released Muse Spark Open on October 15, 2026.

Sources: [Creator release announcement](<https://research.meta.ai/blog/muse-spark-open>)

## Sources and related links

Prices, limits, capabilities, endpoints, and availability come from the public Impossibl catalog. Creator documentation supplies descriptions and release dates where available; it does not override the published API offering.

- [Public Impossibl model catalog](<https://api.impossibl.com/v1/models>)
- [Full model directory as JSON](<https://impossibl.com/models.json>)
- [Muse Spark Open: model details and pricing](<https://impossibl.com/meta/muse-spark-open>)
- [Model directory](<https://impossibl.com/models>)
- [Meta AI Research: Introducing Muse Spark Open](<https://research.meta.ai/blog/muse-spark-open>)
- [OpenRouter: Muse Spark Open](<https://openrouter.ai/meta/muse-spark-open>)
- [Fireworks AI model API: muse-spark-open](<https://api.fireworks.ai/v1/accounts/fireworks/models/muse-spark-open>)
