To address the challenges of developing and maintaining large-scale software codebases, Meta has recently launched a new terminal AI programming agent, Muse Code, which is now available in beta testing. Developed under the leadership of Alexandr Wang, Meta's head of artificial intelligence, the tool is designed to handle complete software engineering tasks—from planning changes and writing code to verifying results—after a single-command installation. Built on Meta's latest AI model, Muse Spark 1.2, Muse Code's core advantage lies in its significant cost-effectiveness compared to competitors. Developers can use the service on a pay-as-you-go basis, with pricing set at $1.25 per million input tokens and $4.25 per million output tokens. Additionally, Muse Code offers a Contributor Version, allowing users to access the service at a significantly lower cost—over 10 times cheaper than the pay-as-you-go model—provided they actively agree to help improve the model. Meta also announced that it accepts zero-data retention requests, ensuring that developers' data will not be used for model improvement and meeting strict corporate data security requirements.