Meta's new Muse Code offers affordable AI coding, but struggles with visuals and lacks key features compared to rivals.
Meta introduces Muse Code
Meta has launched Muse Code, a new code agent aimed at professionals, with a starting price of $5 per month. The tool is part of the company's effort to regain a presence in the generative AI market following the popularity of its Llama series of large language models.Muse Code's performance compared to others
According to one report, Muse Code performs similarly to Claude Code in detecting bugs but falls short in visual output, with its design described as outdated. This suggests that while it is effective in some coding tasks, it may not yet match the visual capabilities of its competitors.Features and limitations of Muse Code
Muse Code includes features such as goal setting, plan mode, parallel sub-agents, and multi-agent workflows, among others. However, it lacks an IDE extension, desktop or mobile apps, and cloud agents, which could limit its usability for some developers.Pricing and model options
The pricing for Muse Code ranges from $5 to $200 per month. The lower-tier plans do not include Muse Wonder Full, the most powerful model for coding and reasoning, which may affect the tool’s overall effectiveness for complex tasks.Testing Muse Code's capabilities
In a test, Muse Code was asked to create a futuristic landing page, resulting in a basic design with animations similar to an open-source Chinese model from six months ago. The visual output was criticized for being outdated and lacking realism, raising questions about its current capabilities in design-related tasks.Muse Code's performance in benchmarks
In benchmark tests, Muse Code performed slightly behind Gemini 4 with Antigravity but matched Claude Opus 5.5 and Gemini 4 in terms of token pricing. However, the exact cost of using Muse Code remains unclear due to the billing structure of Muse Wonder Full, which is not based on per-token charges.