WHY THIS MATTERS
Meta launched Muse Code, its first AI coding agent, in public beta on August 5, 2026, for macOS and Linux. Built on the new Muse Spark 1.2 model, the terminal based tool plans, writes, and reviews whole engineering tasks, with pay as you go pricing and a cheaper contributor tier in exchange for training data. The launch puts Meta in direct competition with Anthropic's Claude Code and OpenAI's Codex.In this article
Meta Launches Muse Code, Its First AI Coding Agent
Meta announced the public beta launch of Muse Code on August 5, 2026, its first AI agent built specifically for software development. The tool runs from the terminal and is currently available in beta for macOS and Linux, with no confirmation yet on support for other platforms. Installation takes a single terminal command, with no extra setup required.
An Agent Built for Full Engineering Tasks
Unlike assistants that just suggest code snippets, Muse Code is designed to handle whole engineering tasks across large codebases: it plans the change, writes the code, and checks the result before handing it back. The tool also supports multiple agents working on the same task at once, with implementation running in parallel while reviewer agents watch and check the work in the background.
Muse Spark 1.2: The Model Behind the Agent
Muse Code runs on Muse Spark 1.2, a coding focused AI model that Meta co trained alongside the agent itself so the two are designed to work together. According to the company, Muse Spark 1.2 improves on the earlier Muse Spark 1.1 in code generation, complex debugging, and codebase comprehension, after Meta scaled up coding specific training compute and widened the diversity of its training environments.
Pay As You Go Pricing and the Contributor Tier
Muse Code is priced on a pay as you go basis, at $1.25 per million input tokens and $4.25 per million output tokens. Meta offers two pricing tiers: a standard rate and a lower cost contributor tier, where users get a significantly reduced price in exchange for letting Meta use their prompts and completions to further train its AI models.
A Direct Challenge to Claude Code and Codex
With Muse Code, Meta becomes a direct competitor to Anthropic's Claude Code and OpenAI's Codex in the AI coding agent space. According to reporting from Forbes, TechCrunch, and CNBC, the launch underscores the intensifying race among major tech companies for a foothold in developers' daily workflows, an area seen as strategically important for enterprise adoption of generative AI.
Why It Matters for Brazilian Agencies and SMBs
Brazilian marketing and customer service automation agencies do not only sell automation to clients, they also build internal tools, integrations, and scripts to run their own operations, increasingly relying on AI coding agents for that day to day work. A new, price competitive entrant like Muse Code widens the field beyond Claude Code and Codex, adding pressure on pricing across the board. The contributor tier, which trades a discount for training data access, deserves particular scrutiny: agencies handling client code and sensitive integrations need to weigh carefully whether that trade off makes sense, or whether paying the standard rate to keep tighter control over the prompts and code they submit is the safer choice.