# Muse Spark 1.2 Contributor

The same Muse Spark 1.2 model on a data-sharing arrangement. Muse Spark 1.2 Contributor costs a fraction of the standard tier because your inputs and outputs are used to train Meta's models, which makes this an eligibility question before it is a pricing one.

- **Model ID:** `meta/muse-spark-1.2-contributor`
- **Type:** chat
- **Providers:** meta
- **Context window:** 1,048,576
- **Maximum output tokens:** 1,048,576
- **Pricing:** $0.10/1M input tokens, $0.2/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/muse-spark-1.2-contributor

## 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-1.2-contributor',
  prompt: 'Why is the sky blue?'
})
```

## About

Muse Spark 1.2 Contributor is a pricing tier rather than a separate model. The weights, the context window of 1.0M tokens, and every capability match Muse Spark 1.2: multimodal input across text, images, video, audio, and PDFs, multi-agent operation as either a planning agent or a parallel subagent, structured output, parallel function calling, and configurable reasoning effort.

What differs is the commercial arrangement. Meta charges substantially less per token, and in exchange uses your inputs and outputs to train and improve its models. Meta has published no separate capability claim for this tier, because there is nothing different to claim.

Decide eligibility before you compare prices. Work that suits this tier is public, synthetic, permissively licensed, or explicitly approved for sharing. Client repositories, personal data, secrets, unreleased product logic, and material under NDA belong on the standard tier.

The scope is wider than the prompt you write. An agent can read terminal output, generated files, and neighbouring directories, so what reaches Meta may exceed what a developer consciously submits. Scope the working directory accordingly.

You can integrate Muse Spark 1.2 Contributor through AI SDK, Chat Completions API, Responses API, Messages API, or other API formats, from TypeScript or Python.

## What to consider

Read the terms in your Meta developer account before routing traffic here. Public documentation on retention period, deletion process, and regional eligibility is thin, and the terms in your logged-in account take precedence over any summary, including this page.

An agent's reach is wider than its prompt. Terminal output, generated files, and neighbouring directories can all enter the context that gets shared, so a developer who intends to submit one file may submit considerably more. Scope the working directory before you point an agent at this tier.

The saving is real and large, which is exactly why the decision should not be made on price. Standard Muse Spark 1.2 exists for everything that fails the eligibility test, and mixing the two per task is a reasonable pattern: the model can be switched mid-session without losing context.

## When to use

### Best For

- **Public and open source code** where sharing inputs carries no confidentiality risk
- **Synthetic and generated data** produced specifically for the workload
- **High-volume experimentation** where the cost difference changes what is affordable
- **Explicitly approved projects** cleared for data sharing in advance

### Consider Alternatives When

- **Client or customer code** is involved, where standard Muse Spark 1.2 is the right tier
- **Secrets or personal data** could enter context, including from files an agent reads on its own
- **NDA or unreleased work** is in scope, regardless of the price difference
- **Retention terms must be documented** before use, since public documentation is limited

## Best for

- **Public And Open Source Code:** Sharing inputs carries no confidentiality risk
- **Synthetic Data:** Generated specifically for the workload
- **High-Volume Experimentation:** Where cost decides what is affordable to try
- **Pre-Approved Projects:** Cleared for data sharing in advance

## Consider alternatives

- **Client Or Customer Code:** Standard Muse Spark 1.2 is the correct tier
- **Secrets And Personal Data:** An agent may read files you did not intend to submit
- **NDA Or Unreleased Work:** The price difference does not change eligibility
- **Undocumented Retention Needs:** Public terms on retention and deletion are limited

## Frequently asked questions

### Is Muse Spark 1.2 Contributor a different model from Muse Spark 1.2?

No. Same weights, same capabilities, same context window of 1.0M tokens. Only the commercial and data-use arrangement differs.

### What is the tradeoff for the lower price?

Meta uses your inputs and outputs to train and improve its models. That makes it an eligibility question about your data rather than a pricing question.

### What work is appropriate for Muse Spark 1.2 Contributor?

Public, synthetic, permissively licensed, or explicitly approved material. Keep client repositories, personal data, secrets, unreleased product logic, and NDA material on the standard tier.

### Could an agent share more than I intend?

Yes. An agent can read terminal output, generated files, and neighbouring directories, so context can include material a developer never consciously submitted. Scope the working directory before pointing an agent here.

### Where are the exact retention and deletion terms?

In your Meta developer account, which takes precedence over any summary. Public documentation on retention period, deletion process, and regional eligibility is limited.

### Can I use both tiers in one project?

Yes. The model can be switched mid-session without losing context, so routing eligible tasks here and everything else to standard Muse Spark 1.2 is a workable pattern.

### Does Muse Spark 1.2 Contributor support Zero Data Retention?

Zero Data Retention is not currently available for this model. Zero Data Retention is offered on a per-provider basis and is a separate matter from this tier's training arrangement. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

## Links

- [Model page](https://vercel.com/ai-gateway/models/muse-spark-1.2-contributor)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://developer.meta.com/ai/models/muse-spark/)