Meta says its new AI model is ready to compete on coding
TL;DR
Meta is opening Muse Spark 1.1 to developers through the new Meta Model API, letting the model plug into AI coding tools. Meta says the model can detect and fix more complex bugs, support end-to-end agentic workflows, and work better across multi-agent systems. Muse Spark 1.1 is available in Thinking mode in the Meta AI app and website. The API launches as a public preview for US developers.
Nauti's Take
This looks like a serious move, but it also carries plenty of PR pressure. Meta has reach, capital, and distribution, yet coding models are judged by benchmarks, tool integration, latency, price, and failure rate.
Free API credits reduce friction, but they do not prove production usefulness. Muse Spark 1.1 becomes more than catch-up rhetoric only if it performs clearly better inside real coding-agent workflows.
Briefingshow
Coding agents are one of the toughest proving grounds for new AI models because they must produce working changes, not just plausible text. Meta is trying to turn Meta AI into a developer platform. The real test is whether Muse Spark 1.1 is reliable enough inside actual IDE and agent workflows.
Sources