Muse Spark 1.2 Explained: Meta's Model, Price and Modes

In short: Muse Spark 1.2 is an AI model from Meta, released on 5 August 2026 and available through Meta Model API and Muse Code. It costs $1.25 per million input tokens and $4.25 per million output in Standard mode, or $0.10 and $0.20 in Contributor mode, where Meta may use your content to train future models. It reads text, images, video, audio and PDFs, with a 1,048,576-token context window. Meta's newer version is Muse Spark 1.3.
What is Muse Spark 1.2?
Muse Spark 1.2 is a large language model in Meta's Muse Spark line, built for agentic and coding work. Meta's models page describes Muse Spark as its model for "agentic and coding work, such as multi-step tool loops, software engineering assistants, and long-context reasoning". The API model ID is muse-spark-1.2.
Muse Spark comes in three versions - 1.1 (the original), 1.2 (the previous version) and 1.3 (the latest) - all with the same inputs and context window. For how models like this work, see how LLMs work.
How does Muse Spark relate to Llama?
Meta's earlier Llama models were released as open weights; the hosted Muse Spark versions are served through Meta's own API. Meta's documentation lists six model lines - Muse Wonder, Muse Spark, Muse Image, Muse Voice Transcribe, Segment Anything Model and Muse Glimmer - and notes that Muse Spark also has an open-weight release, Muse Spark Open, alongside the open-weight Muse Glimmer.
How much does Muse Spark 1.2 cost?
Meta prices Muse Spark in two modes: Standard, where your content is not used for training, and Contributor, which is heavily discounted in exchange for training use. From Meta's pricing page, the same for Muse Spark 1.3, 1.2 and 1.1:
- No long-context premium - the rate is the same whether the context window is nearly empty or nearly full.
- Web search grounding costs $2.50 per 1,000 search queries, on top of the token cost.
- Contributor mode is chosen by model ID:
muse-spark-1.2-contributorinstead ofmuse-spark-1.2.
What is Contributor mode?
Contributor mode is a discounted way to use Meta's models in which Meta may use your content to train future models. Meta defines that content as the prompts, code, files and context you send, plus the model's responses. In Standard mode, Meta says your content is not used to train its models. Every model on Meta's pricing page is available in Standard mode.
What are Muse Spark 1.2's specs?
| Spec | Muse Spark 1.2 |
|---|---|
| Developer | Meta |
| Released | August 5, 2026 |
| Model ID | muse-spark-1.2 (Contributor: muse-spark-1.2-contributor) |
| Context window | 1,048,576 tokens |
| Input | Text, images, video, audio and PDF |
| Output | Text |
| Features | Tool calling, structured output, image, video and audio understanding, search grounding |
| APIs | Responses, Chat Completions and Messages |
| Rate limits | 3,000 requests and 4,000,000 tokens per minute (Standard) |
| Where to use it | Meta Model API (pay-as-you-go) and Muse Code |
How does Muse Spark 1.2 compare with Muse Spark 1.3?
Both have the same prices, inputs and context window; Meta describes 1.3 as tuned for agentic workflows with improved coding over 1.2. Muse Spark 1.3 supports all reasoning effort levels, including max. Meta's documentation notes that audio understanding in Muse Spark 1.3 is not fully supported, and points to Muse Spark 1.2, or Muse Voice Transcribe, for audio.
How does Muse Spark 1.2 rank?
On LMArena's text leaderboard Muse Spark 1.2 has 1492 Elo (rank 13, xhigh effort). Muse Spark 1.3 has 1495 Elo (rank 11, max effort). The full ranked table is on our benchmarks page.
Words to know
- Standard mode - normal pricing; your content is not used to train Meta's models.
- Contributor mode - discounted pricing in exchange for Meta training on your content.
- Cached input - repeated input billed at a lower rate.
- Search grounding - letting the model use web search results in its answer.
- Open weights - model files published for anyone to download and run.
Related guides
Frequently asked questions
What is Muse Spark 1.2?+
Muse Spark 1.2 is a Meta AI model for agentic and coding work, released on 5 August 2026. It accepts text, images, video, audio and PDF input, returns text, and has a 1,048,576-token context window.
How much does Muse Spark 1.2 cost?+
In Standard mode it costs $1.25 per million input tokens and $4.25 per million output tokens. In Contributor mode it costs $0.10 and $0.20, in exchange for Meta being allowed to use your content to train future models.
What is the difference between Muse Spark 1.2 and 1.3?+
They share prices, inputs and the context window. Meta describes Muse Spark 1.3 as tuned for agentic workflows with improved coding, and recommends Muse Spark 1.2 for audio because audio understanding in 1.3 is not fully supported.
Is Muse Spark open source?+
The hosted Muse Spark versions run on Meta Model API. Meta also lists an open-weight release, Muse Spark Open, that can run on your own or a provider's hardware.
Where can I use Muse Spark 1.2?+
Through Meta Model API on pay-as-you-go billing, and in Muse Code, Meta's coding agent, on pay-as-you-go or a subscription.
Enjoyed this? Get more in your inbox.
Up to ten of the week's AI news stories, by email once a week.