Muse Code and Muse Spark 1.2

Meta is rolling out Muse Code, a new assistant that can help write, plan, and check software work, powered by its upgraded model, Muse Spark 1.2. On paper, this is a big swing: the bot is supposed to handle long, messy coding jobs, remember what it was doing after crashes, and keep working in the background like a tiny team of digital interns. Meta is also bragging that it can grind through giant tasks for hours, even using it to improve graphics-chip code over a full day of nonstop work.

But in the comments, the real action is less “wow” and more “wait a minute…”. One of the loudest reactions was pure suspicion: why did version 1.1 arrive just weeks ago, only for 1.2 to show up almost immediately? That led to the hottest theory of the thread — that this is basically a do-over launch after the earlier release got overshadowed. Another mini-drama erupted over comparisons, with people asking why Meta measured itself against some rivals but seemingly skipped the one commenter really wanted to see, GPT-5.6-sol. Translation for non-tech readers: people think the company may have picked safer opponents.

Then came the privacy panic. One commenter flat-out asked why login is required at all, and whether anyone should trust Meta with access to private code. That vibe — cool demo, but who’s watching? — hung over the whole discussion. The lighter side? A few people zeroed in on the actually exciting bit: broader availability outside the U.S. For some readers, that was the biggest news of all. In classic internet fashion, the launch turned into a mix of hype, side-eye, and a full-on trust issue.