Rank of 27161from #15

Known for

Long, multi-agent coding sessions on Meta's own agent, or through OpenRouter.

Meta's coding-tuned model, shipped alongside the Muse Code agent on 5 August 2026 and superseded by 1.3 on 2 September. Trained with its own harness for reliable tool calling.

What it does7

  • 1M-token context
  • Text, image, video, audio and PDF in
  • Co-trained with the Muse Code harness
  • Context compaction for long runs
  • Async background subagents
  • Replay-exact event log, restart-safe
  • Contributor tier at $0.10 / $0.20 in exchange for training use

From Muse Spark 1.2’s own site. Not used for ranking.

43index score · LLMs

How well does it score?

x
bottom 24%
x
bottom 46%
x
top 50%
x
1.25 / 4.25 · top 22%

Compared with everyone else on this board.

—votes · crowd
Reading votes…

Votes don’t change the score.

Buy me a coffee