Announcing MCP Toolbox Java SDK v1.0: Agentic data access for the enterprise
cloud.google.com·6d ago
Cloud & Infrastructure
The New Stack·Sep 3·5 min read
“Google was ahead only a few hours”: Muse Spark 1.3 edges out Gemini as Meta claims its biggest coding leap yet
TL;DR
Meta has launched Muse Code 1.3, a coding agent that reportedly surpasses Google's Gemini in performance, marking a significant advancement in AI coding tools.
✦ Why It Matters
Explore Muse Code 1.3 today to enhance your coding efficiency and stay competitive in AI-driven development.
Key Takeaways
- Evaluate Muse Spark 1.3 for your coding tasks to leverage its performance.
- Monitor the upcoming open-weight release for independent model deployment.
- Compare performance metrics with competitors like Gemini 3.8 Flash.
- Consider subscription plans for Muse Code to enhance your development workflow.
Full Summary
Meta has officially launched Muse Spark 1.3, its latest low-cost reasoning model, which is now available through Muse Code and the Meta Model API. This version boasts significant improvements in coding and agentic tasks, with Meta CEO Mark Zuckerberg claiming it delivers 'frontier performance almost too cheap to meter.'
Muse Spark 1.3 scored 75.4% on the DeepSWE coding benchmark and 98.1% on the MRCR long-context test, outperforming Google's recently released Gemini 3.8 Flash. The model is also noted for its cost-effectiveness, priced at approximately $0.55 per task, making it the most economical option among its peers.
However, the results were generated under a limited preview mode, which may not reflect the typical performance available to developers. Meta is also preparing to release open-weight versions of Spark, allowing developers to run the model independently, though licensing details remain unclear.