Muse Spark 1.2 Explained

Muse Spark 1.2 is the model Meta released on 2026-08-05 to power its new Muse Code coding agent. It follows Muse Spark 1.1 in Meta's single Muse Spark line.

This guide covers what Muse Spark 1.2 is, how it differs from 1.1, the two-tier pricing table, and where teams should place it in their stack.

Every price and feature below is quoted from the launch announcement. Anything Meta has not published (rate limits, context window, benchmarks) is called out so you can verify at the vendor page.

What Is Muse Spark 1.2?

Muse Spark 1.2 is Meta's coding-grade language model released on 2026-08-05 to back the Muse Code CLI agent.

Meta ships a single Muse Spark line, so 1.2 is a direct successor to Muse Spark 1.1. There is no sibling routing decision inside Meta's own catalog.

The model is available through Muse Code and the Meta developer platform. Meta has not published rate limits, context window, or output-token caps at launch — verify at https://developer.meta.com/ai/products/muse-code/.

• Vendor: Meta
• Launch date: 2026-08-05
• Backs: Muse Code CLI coding agent
• Predecessor: Muse Spark 1.1

Two-Tier Pricing: Standard vs Contributor

Muse Spark 1.2 ships in two price tiers that trade data-use for cost. Standard is the default privacy-preserving tier; Contributor is roughly 12x cheaper in exchange for Meta using your prompts and outputs to improve their products.

Standard costs $1.25 per million input tokens, $0.15 per million cached input tokens, and $4.25 per million output tokens.

Contributor costs $0.10 per million input tokens, $0.002 per million cached input tokens, and $0.20 per million output tokens. Meta uses Contributor prompts and outputs to improve their products.

• Standard: $1.25 in / $0.15 cached / $4.25 out per M tokens (no data-use)
• Contributor: $0.10 in / $0.002 cached / $0.20 out per M tokens (data-use YES)
• Standard is the default privacy-preserving tier

How Muse Spark 1.2 Differs From 1.1

Muse Spark 1.2 is tuned for long-horizon, multi-agentic coding workflows, which is how Meta positions Muse Code.

Muse Spark 1.1 was Meta's general multimodal reasoning model with a 1M-token context window and agentic tool use. 1.2 keeps the coding-agent focus and layers on the two-tier data-use pricing structure.

Meta has not published side-by-side benchmark scores at launch. Teams already on 1.1 should A/B test their own workloads before switching — verify capability deltas at https://developer.meta.com/ai/products/muse-code/.

• 1.2 is optimized for long-horizon multi-agent coding workflows
• 1.2 introduces the Contributor discount tier
• 1.1 remains the reference for general multimodal reasoning

Where Muse Spark 1.2 Fits in Your Stack

Muse Spark 1.2 fits teams that want a CLI-native coding agent with an event log they can replay, without paying Claude Fable 5 or GPT-5.6 prices for every commit.

The Standard tier is where regulated teams and client-facing shops should live, because prompts and outputs are not used for training. When we run our own competitor-monitor routine across the layer3 portfolio we keep client code on privacy-preserving tiers for the same reason.

The Contributor tier is the right fit for open-source contributors, personal side projects, and non-sensitive prototypes where the 12x cost cut matters more than data-use terms.

• Standard for client work and regulated data
• Contributor for OSS and non-sensitive prototypes
• Pair with Muse Code for the multi-agent, event-log workflow

How to use Muse Spark 1.2

You do not host Muse Spark 1.2 yourself — you use it through a tool, so "getting started" really means choosing the right one.

The fastest way to put Muse Spark 1.2 to work day to day is inside an AI IDE, and Cursor is the most popular — it supports it directly, so you can be working in minutes. Prefer a different editor? Windsurf, Zed, and GitHub Copilot drive these models too.