Meta joins independent programming race and lowers prices

Meta launched its new programming tool, "Muse Code," on August 5th, with the aim of enhancing competition in the AI coding tools market and competing against models like "Codex" and "Claude Code." The tool is based on the "Muse Spark 1.2" model and operates via command line on macOS and Linux systems. It features capabilities such as managing long-term programming projects, planning modifications, writing and testing code independently, and maintaining an event log that allows resuming work after outages. Meta seeks to develop more autonomous programming tools, enabling the agent to perform multiple tasks, plan, and execute tests, thereby reducing human intervention. "Muse Code" is billed on a usage-based model, with prices starting at $1.25 per million tokens. The company believes these prices are significantly lower than competitors', which could attract individual developers and startups, despite concerns over code security and data privacy. Initial results indicate improved performance compared to previous versions, but the level of risk associated with the agent's ability to read and modify files raises security concerns. In summary, "Muse Code" aims to transform the concept of smart programming tools from mere assistants into fully autonomous agents capable of executing entire tasks, raising privacy and security challenges in the process.

Size: 120%Text Size

Notice: This Is an AI-Generated Summary

Display The Full Article