Meta Releases Muse Code, Its First AI Coding Agent, and Competes With Claude Code and Codex on Price

The terminal agent runs on Muse Spark 1.2, trails Anthropic's Opus 5 on Meta's own coding benchmarks and offers a cheaper tier for developers who share data to improve the model.
Meta Releases Muse Code, Its First AI Coding Agent, and Competes With Claude Code and Codex on Price
Meta Releases Muse Code, Its First AI Coding Agent, and Competes With Claude Code and Codex on Price

Meta has released Muse Code, its first AI coding agent, in a preview aimed at the developers who now use Anthropic’s Claude Code and OpenAI’s Codex. The agent runs in the terminal on Muse Spark 1.2, a coding-focused update to Meta’s model that the company trained alongside it, CNBC reported.

It is the latest release from Alexandr Wang, who leads Meta Superintelligence Labs. “You can install it with one command and then use it to take on complete software engineering tasks across a wide variety of use cases, planning changes, writing code, validating the results,” Wang told CNBC in an interview Wednesday. He declined to share usage numbers for the Muse Spark models but said “adoption has been exciting and strong.”

Meta’s pitch rests on how the agent runs long jobs. Muse Code can coordinate several persistent subagents per task and logs every model call, tool run, approval and edit to a local event log, Decrypt reported. Meta said that makes the runtime “replay-exact and restart-safe: after a crash, the agent can resume precisely where it stopped.” It ships with /plan, /grill and /goal commands, and in one demo it turned an mp4 fly-through of a house into a website with booking features. Meta also said the agent “iteratively optimized GPU kernels over 1,000+ tool calls (up to 24 hours) on Nvidia Hopper GPUs.”

On Meta’s own charts, Muse Spark 1.2 with Muse Code scored 82.9% on Terminal-Bench 2.1, behind Claude Code on Opus 5 at 86.7% but ahead of GPT-5.6 Terra on Codex at 81.8% and Grok Build at 81.6%. On DeepSWE 1.1 it scored 59.3%, against 65.0% for Opus 5 and 64.8% for Codex. Meta’s internal coding benchmark showed 70.6% for Muse and 79.4% for Opus 5.

Wang said Meta is competing on price rather than capability. Pay-as-you-go rates match Muse Spark 1.1 at $1.25 per million input tokens and $4.25 per million output tokens. A contributor tier is “more than 10 times cheaper than even the pay-as-you-go tier,” he said, but developers on it must “opt-in to help improve the model.” Meta is also starting to accept requests for zero-data retention, which Wang called “a big enterprise feature that is important for folks.”

Developers can sign up and pay on the same Meta developer page that hosts the Muse Spark API, and Meta’s newer model will also be offered on OpenRouter. The agent installs with a single curl script from dev.meta.ai.

Subscribe for Updates

AIDeveloper Newsletter

The AIDeveloper Email Newsletter is your essential resource for the latest in AI development. Whether you're building machine learning models or integrating AI solutions, this newsletter keeps you ahead of the curve.

By signing up for our newsletter you agree to receive content related to ientry.com / webpronews.com and our affiliate partners. For additional information refer to our terms of service.

Notice an error?

Help us improve our content by reporting any issues you find.

Get the WebProNews newsletter delivered to your inbox

Get the free daily newsletter read by decision makers

Subscribe
Advertise with Us

Ready to get started?

Get our media kit

Advertise with Us