Meta has launched Muse Code in beta, a terminal-based coding agent powered by the new Muse Spark 1.2 model, putting the company directly into the developer-agent market currently defined by Anthropic’s Claude Code and OpenAI’s Codex. The practical catch for Windows shops is immediate: Muse Code’s current install path is for macOS, Linux, and Windows Subsystem for Linux, rather than a native PowerShell or Windows terminal release.
Business Insider first reported the launch’s central commercial proposition on August 5: Meta is selling Muse Spark 1.2 API access below the rates commonly associated with high-end coding models, while positioning Muse Code as an agent that can write code, run tests, and validate results across complex projects. The Tech Buzz also reported the launch, but its claim that Meta had not detailed pricing, availability, or technical specifications was already out of date when published; Meta’s published pricing and the beta’s platform limitations were available through launch coverage.
The important development is not simply that Meta has another code model. Muse Code is Meta’s first attempt to ship the entire coding-agent stack: a command-line harness, persistent worker agents, worktree isolation, long-context model access, and a pricing model that explicitly discounts development data Meta can reuse.
For developers, that makes Muse Code a real Claude Code and Codex alternative worth testing. For administrators and organizations with Windows-heavy, proprietary repositories, it also creates a more complicated risk and deployment decision than the launch messaging suggests.
Meta says Muse Code can perform whole-repository engineering work rather than simply autocomplete a function or answer a coding question. Its design centers on a main agent supported by specialized background agents that remain active through a session, instead of being created and discarded for each task.
Those background agents can investigate a codebase, plan changes, run tools, and report findings back to the main process. For larger jobs, Muse Code can hand work to parallel sub-agents in isolated Git worktrees, an approach intended to prevent simultaneous agents from modifying the same files in a developer’s checked-out branch.
Meta has also built an append-only local event log into the runtime. The company says that log records model calls, tool use, edits, approvals, and other state changes, allowing a crashed or interrupted session to resume rather than rebuilding its context from scratch. That is a meaningful feature for a tool meant to run over many hours: the operational problem with long-running coding agents is often not model capability but whether a failed terminal session loses the state needed to finish safely.
The beta includes workflow commands intended to put a checkpoint before the most consequential work. Meta describes
Meta demonstrated the agent making more than 1,000 tool calls during GPU-kernel optimization tasks that ran for as long as 24 hours. That is evidence that the harness can sustain a lengthy workload under controlled conditions. It is not evidence that Muse Code will safely refactor a company’s mixed-age .NET, PowerShell, JavaScript, and infrastructure-as-code estate without human review.
That puts Muse Code behind Claude Code on platform coverage. Anthropic’s current Claude Code documentation includes native Windows installation through PowerShell and WinGet, alongside macOS, Linux, and WSL options. OpenAI’s Codex is also available in developer terminal workflows that do not require a Linux compatibility layer as the default Windows path.
WSL is a practical environment for many developers, especially those already using Ubuntu, Docker, dev containers, Git, and VS Code Remote - WSL. But it is not a transparent substitute for native Windows support in managed fleets. Repository locations, credentials, corporate proxies, endpoint protection, Git configuration, line-ending behavior, filesystem performance, and access to Windows-native build tools all need to be tested rather than assumed.
The worktree model adds another operational consideration. Isolating agents in worktrees can reduce direct file collisions, but it also means organizations should expect additional repository copies and build artifacts on developer machines. On a WSL setup that crosses between the Linux filesystem and mounted Windows drives, that can make performance and path discipline more important than it would be for a simple chat-based assistant.
For a Windows development team, the sensible initial use case is a disposable WSL-hosted clone with constrained credentials and no deployment access. Let Muse Code inspect, plan, make a branch, and run local tests. Do not begin by giving it a production-connected shell, broad cloud credentials, or access to the only working copy of a legacy application.
That distinction is being glossed over in the competitive framing. Anthropic and OpenAI bundle their coding products into plans with usage limits, while Meta is quoting metered model costs. A developer who sends a few prompts a day and a team that keeps a long-running agent indexed against a large repository will produce radically different bills. Cached-context pricing helps when the same repository material is reused, but it does not answer the budget question without real telemetry from a working project.
Meta also offers a sharply discounted “contributor” tier under which user activity may be used to improve Meta’s products. That is the launch’s most consequential commercial term. The cheapest Muse Spark option is not simply a lower-performance plan; it is a data-sharing choice.
For hobby projects, prototypes, public open-source work, or carefully sanitized test repositories, that may be an acceptable trade. It could allow small teams to experiment with long-context, agentic coding at a lower cost than competitors. For source code containing customer logic, trade secrets, unreleased product plans, credentials accidentally left in configuration files, or regulated data, it should be treated as disqualifying unless the organization has separately cleared that use.
Meta’s standard paid tier is presented as the alternative for customers that do not want their activity used for product improvement. But the launch coverage does not establish the full set of enterprise controls a security team will ask about: retention periods, contractual data-processing terms, tenant isolation, audit logging, identity integration, administrative controls, regional processing, and the exact handling of tool outputs. Those omissions do not prove the controls are absent. They do mean a public beta announcement is not enough to approve the agent for sensitive repositories.
Meta has also highlighted Terminal-Bench 2.1 results for the model. Benchmarks are useful indicators of whether an agent can interact with shells and development tools under repeatable conditions, but they do not measure the things that make an enterprise rollout succeed: whether the tool honors local conventions, makes understandable commits, avoids destructive changes, works through a corporate proxy, and stops when uncertainty is high.
The biggest comparative disadvantage is maturity, not necessarily model intelligence. Claude Code and Codex have built user habits, editor integrations, configuration files, permissions models, and accumulated knowledge around failure modes. A coding agent is not adopted merely because it writes a good patch; it is adopted because a team can make its behavior predictable enough to put it inside an existing review and release process.
Meta has entered that race with an intriguing technical architecture and a price lever that rivals may have trouble matching. It has not yet shown that it can provide the surrounding administration, Windows support, and trust controls needed to move from experimental developer tooling to broad enterprise use.
Muse Code is available globally in beta, but for Windows organizations that should be read as available to evaluate through WSL, not ready to standardize across native Windows development environments.
The important development is not simply that Meta has another code model. Muse Code is Meta’s first attempt to ship the entire coding-agent stack: a command-line harness, persistent worker agents, worktree isolation, long-context model access, and a pricing model that explicitly discounts development data Meta can reuse.
For developers, that makes Muse Code a real Claude Code and Codex alternative worth testing. For administrators and organizations with Windows-heavy, proprietary repositories, it also creates a more complicated risk and deployment decision than the launch messaging suggests.
Muse Code is designed to keep working after the first prompt
Meta says Muse Code can perform whole-repository engineering work rather than simply autocomplete a function or answer a coding question. Its design centers on a main agent supported by specialized background agents that remain active through a session, instead of being created and discarded for each task.Those background agents can investigate a codebase, plan changes, run tools, and report findings back to the main process. For larger jobs, Muse Code can hand work to parallel sub-agents in isolated Git worktrees, an approach intended to prevent simultaneous agents from modifying the same files in a developer’s checked-out branch.
Meta has also built an append-only local event log into the runtime. The company says that log records model calls, tool use, edits, approvals, and other state changes, allowing a crashed or interrupted session to resume rather than rebuilding its context from scratch. That is a meaningful feature for a tool meant to run over many hours: the operational problem with long-running coding agents is often not model capability but whether a failed terminal session loses the state needed to finish safely.
The beta includes workflow commands intended to put a checkpoint before the most consequential work. Meta describes
/plan as an approval-gated planning step, /grill as a way to stress-test a plan, and /goal as a mechanism for keeping an agent directed at a specified outcome. The presence of those controls is useful, but it should not be mistaken for a review system. An approval gate is only as strong as the person reading the plan and the repository permissions granted to the process.Meta demonstrated the agent making more than 1,000 tool calls during GPU-kernel optimization tasks that ran for as long as 24 hours. That is evidence that the harness can sustain a lengthy workload under controlled conditions. It is not evidence that Muse Code will safely refactor a company’s mixed-age .NET, PowerShell, JavaScript, and infrastructure-as-code estate without human review.
The Windows story is WSL, not native Windows
For Windows users, the first deployment detail matters more than the benchmark numbers: Muse Code is presently delivered for macOS and Linux, with WSL serving as the route for Windows machines. Meta has not announced a native Windows installer, a PowerShell-first workflow, or a Windows Terminal-specific build.That puts Muse Code behind Claude Code on platform coverage. Anthropic’s current Claude Code documentation includes native Windows installation through PowerShell and WinGet, alongside macOS, Linux, and WSL options. OpenAI’s Codex is also available in developer terminal workflows that do not require a Linux compatibility layer as the default Windows path.
WSL is a practical environment for many developers, especially those already using Ubuntu, Docker, dev containers, Git, and VS Code Remote - WSL. But it is not a transparent substitute for native Windows support in managed fleets. Repository locations, credentials, corporate proxies, endpoint protection, Git configuration, line-ending behavior, filesystem performance, and access to Windows-native build tools all need to be tested rather than assumed.
The worktree model adds another operational consideration. Isolating agents in worktrees can reduce direct file collisions, but it also means organizations should expect additional repository copies and build artifacts on developer machines. On a WSL setup that crosses between the Linux filesystem and mounted Windows drives, that can make performance and path discipline more important than it would be for a simple chat-based assistant.
For a Windows development team, the sensible initial use case is a disposable WSL-hosted clone with constrained credentials and no deployment access. Let Muse Code inspect, plan, make a branch, and run local tests. Do not begin by giving it a production-connected shell, broad cloud credentials, or access to the only working copy of a legacy application.
Meta’s low-price pitch comes with a data-use fork
Business Insider reports that standard Muse Spark 1.2 access costs $1.25 per million input tokens, $0.15 per million cached-input tokens, and $4.25 per million output tokens. Those are usage prices for the model behind Muse Code, not a clean like-for-like replacement for Claude Code or Codex subscriptions.That distinction is being glossed over in the competitive framing. Anthropic and OpenAI bundle their coding products into plans with usage limits, while Meta is quoting metered model costs. A developer who sends a few prompts a day and a team that keeps a long-running agent indexed against a large repository will produce radically different bills. Cached-context pricing helps when the same repository material is reused, but it does not answer the budget question without real telemetry from a working project.
Meta also offers a sharply discounted “contributor” tier under which user activity may be used to improve Meta’s products. That is the launch’s most consequential commercial term. The cheapest Muse Spark option is not simply a lower-performance plan; it is a data-sharing choice.
For hobby projects, prototypes, public open-source work, or carefully sanitized test repositories, that may be an acceptable trade. It could allow small teams to experiment with long-context, agentic coding at a lower cost than competitors. For source code containing customer logic, trade secrets, unreleased product plans, credentials accidentally left in configuration files, or regulated data, it should be treated as disqualifying unless the organization has separately cleared that use.
Meta’s standard paid tier is presented as the alternative for customers that do not want their activity used for product improvement. But the launch coverage does not establish the full set of enterprise controls a security team will ask about: retention periods, contractual data-processing terms, tenant isolation, audit logging, identity integration, administrative controls, regional processing, and the exact handling of tool outputs. Those omissions do not prove the controls are absent. They do mean a public beta announcement is not enough to approve the agent for sensitive repositories.
Benchmark claims do not settle the production question
Meta says Muse Spark 1.2 was trained alongside Muse Code’s harness, including trajectories covering planning, compaction, sub-agent coordination, repository-scale generation, and longer engineering tasks. Co-training the model and the agent environment can be an advantage: the model is less likely to stumble over its own tool syntax or lose the thread of a multi-step process.Meta has also highlighted Terminal-Bench 2.1 results for the model. Benchmarks are useful indicators of whether an agent can interact with shells and development tools under repeatable conditions, but they do not measure the things that make an enterprise rollout succeed: whether the tool honors local conventions, makes understandable commits, avoids destructive changes, works through a corporate proxy, and stops when uncertainty is high.
The biggest comparative disadvantage is maturity, not necessarily model intelligence. Claude Code and Codex have built user habits, editor integrations, configuration files, permissions models, and accumulated knowledge around failure modes. A coding agent is not adopted merely because it writes a good patch; it is adopted because a team can make its behavior predictable enough to put it inside an existing review and release process.
Meta has entered that race with an intriguing technical architecture and a price lever that rivals may have trouble matching. It has not yet shown that it can provide the surrounding administration, Windows support, and trust controls needed to move from experimental developer tooling to broad enterprise use.
Muse Code is available globally in beta, but for Windows organizations that should be read as available to evaluate through WSL, not ready to standardize across native Windows development environments.
References
- Primary source: Business Insider
Published: 2026-08-05T21:31:45.090000+00:00
Meta Enters Coding Wars With Muse, Taking on Claude and Codex - Business Insider
Meta launched Muse Code, which aims to be a comparable coding agent that's priced lower than competitors like Anthropic's Claude or OpenAI's Codex.www.businessinsider.com - Independent coverage: The Tech Buzz
Published: Wed, 05 Aug 2026 19:46:00 GMT
Meta Launches Muse Code AI Agent to Challenge OpenAI, Anthropic | The Tech Buzz
Meta enters AI coding tools battle with Muse Code, its first autonomous coding agentwww.techbuzz.ai - Related coverage: developersdigest.tech
Meta Launches Muse Spark 1.1: A Closed-Weights Agentic Model with Aggressive Pricing - Developers Digest
Meta's first paid API model arrives with $1.25/M input tokens, 1M context window, and strong tool-use benchmarks. HN debates what it means for the...www.developersdigest.tech - Related coverage: ai.meta.com
- Related coverage: ai.meta.com
- Related coverage: about.fb.com
Introducing Muse Spark: Meta's Most Powerful Model Yet
Muse Spark powers a smarter and faster Meta AI assistant, and will be rolling out to WhatsApp, Instagram, Facebook, Messenger, and AI glasses in the coming weeks.about.fb.com - Related coverage: eweek.com
Meta Launches Muse Spark 1.1: A Lower-Cost AI Model for Coding Agents | eWeek
Meta opens Muse Spark 1.1 to US developers at lower prices, targeting AI coding agents and rivaling OpenAI and Anthropic.
www.eweek.com
- Related coverage: the-decoder.com
Google and Meta race to build personal AI agents as Anthropic and OpenAI pull further ahead
Google and Meta are both internally testing personal AI agents - codenamed "Remy" and "Hatch" - designed to handle everyday tasks on their own, a direct response to the lead built by Anthropic and OpenAI. Google shut down its browser agent project Mariner to focus on the effort. The market is...the-decoder.com - Related coverage: static.businessinsider.com
- Related coverage: axios.com
Meta's Muse Spark agents connects to Google Calendar and Gmail
Meta AI can now connect to Google Calendar and Gmail and handle tasks such as daily briefings, research and redecorating projects.www.axios.com
- Related coverage: axios.com
Meta updates its Spark model, releases developer version
Muse Spark 1.1 promises improvements in coding and handling longer tasks, Meta's Alexandr Wang tells Axios. - Related coverage: tomsguide.com
I tried Muse Spark — here’s what you need to know about Meta’s new AI model | Tom's Guide
Muse Spark, Meta’s newly released proprietary multimodal AI model, owes its development to Meta Superintelligence Lab and promises quality results across its many useswww.tomsguide.com - Related coverage: androidcentral.com
Meta's AI plans look agentic with a potential Instagram bot that shops for you | Android Central
The company's Muse Spark AI might also get an "agentic" boost.www.androidcentral.com - Related coverage: itpro.com
Google's new Jules coding agent is free to use for anyone – and it just got a big update to prevent bad code output | IT Pro
Google publicly released its Jules coding agent earlier this month, and it’s already had a big update aimed at shoring up code quality.www.itpro.com - Related coverage: github.com
MUSE/data/get_evaluation.sh at main · facebookresearch/MUSE · GitHub
A library for Multilingual Unsupervised or Supervised word Embeddings - MUSE/data/get_evaluation.sh at main · facebookresearch/MUSE
github.com
- Related coverage: github.com
GitHub - facebookresearch/MUSE: A library for Multilingual Unsupervised or Supervised word Embeddings · GitHub
A library for Multilingual Unsupervised or Supervised word Embeddings - facebookresearch/MUSE
github.com
- Related coverage: gist.github.com
list of google code project (some are unexportable, have too many issue or already exported) · GitHub
list of google code project (some are unexportable, have too many issue or already exported) - gc-projects.txt
gist.github.com
- Related coverage: support.claude.com
Your first day in Claude Code | Claude Help Center
support.claude.com
- Related coverage: support.claude.com
Claude Code user FAQ | Claude Help Center
- Related coverage: code.visualstudio.com
Terminal Profiles
Visual Studio Code's integrated terminal allows configuring various profiles to make launching various shells easier.code.visualstudio.com - Related coverage: code.visualstudio.com
- Related coverage: investing.com
Meta scraps AI image feature days after launch following privacy backlash By Reuters
Meta scraps AI image feature days after launch following privacy backlashwww.investing.com - Related coverage: developertoolkit.ai
Beta Features and Early Access | Developer Toolkit
How to access, test, and give feedback on Claude Code's beta and research-preview features - agent view, /goal workflows, agent teams, fast mode, and the Figma Dev Mode MCP server.developertoolkit.ai
- Related coverage: marketplace.visualstudio.com
Agent Terminal - Visual Studio Marketplace
Extension for Visual Studio Code - Run a configurable agent/REPL in a terminal in the editor area. Profiles let you choose between multiple CLIs.marketplace.visualstudio.com - Related coverage: marketscreener.com
Meta to put AI chip into production in September as it looks to double computing capacity, memo shows | MarketScreener
Meta Platforms plans to start manufacturing an artificial intelligence chip from September as part of its plan to boost overall computing power to 14 gigawatts next year, showed an internal memo...www.marketscreener.com
- Related coverage: linkedin.com
July 2026: Introducing Meta Glasses With EssilorLuxottica, Generative AI With Muse Image and Muse Video, and Career Lessons From an Engineer-Turned-VP | Meta | 74 comments
Welcome back to Behind the Build! This month we're shipping across hardware, generative AI, and career growth 🚀 From AI-powered glasses to the first models out of Meta Superintelligence Labs (MSL) — here's what our teams built: 𝙄𝙣 𝙩𝙝𝙞𝙨 𝙚𝙙𝙞𝙩𝙞𝙤𝙣: 🕶️ Meta Glasses → New AI glasses with...www.linkedin.com
- Related coverage: theregister.com
Meta admits its first ‘superintelligence’ was too stupid to survive for three days
Pulls AI-powered image tweaker after allowing free-for-allwww.theregister.com - Related coverage: arstechnica.com
Meta's Superintelligence Lab unveils its first public model, Muse Spark - Ars Technica
Meta touts strong benchmarks but admits "performance gaps" in agentic and coding systems.arstechnica.com - Related coverage: vp-land.com
Meta Launches Agentic Muse Image and Previews Muse Video from Superintelligence Labs — VP Land
Meta Superintelligence Labs launched Muse Image, an image model that can call its own coding and web-search tools mid-generation, and opened an early preview of Muse Video, a companwww.vp-land.com - Related coverage: ppc.land
Meta drops Instagram tagging tool, preps Muse Image for advertisers
Muse Image launched July 7, 2026, then lost Instagram tagging three days later after user objections. Advertisers gain Advantage+ creative access within weeks.ppc.land
- Related coverage: investing.com
Meta debuts Muse Spark 1.1 with preview open to developers By Reuters
Meta debuts Muse Spark 1.1 with preview open to developerswww.investing.com - Related coverage: mobileworldlive.com
Meta sparks superintelligence push in... - Mobile World Live
Meta unveiled the first LLM to come out of its lab working on so-called superintelligence, a move it expects to improve AI on its apps.www.mobileworldlive.com
- Related coverage: citizen.org
- Related coverage: twinleafresearch.com
