Muse Wonder
A hosted multimodal model for sustained coding, computer use and personal or professional agents, with a million-token context and persistent-context features. Released through Meta Model API and Muse Code.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why this model scores 84.9
Professional and coding tests support capability 90; sustained-task training supports autonomy 90. Hosted reach gives deployment 80. Family cyber/bio proxies imply misuse 87 despite refusals. Control difficulty 74 balances system safeguards against multi-step errors and incomplete coverage. These are comparative editorial judgments.
Videos about Muse Wonder
Reviewed videos from Artificial Intelligence Videos. These links add context and never change a DoomBench score.
News tied to Muse Wonder
The model score of 84.9 rates this model's risk profile. The overall Doom Index of 69.0 measures the complete temporally weighted evidence record. These values answer different questions.
Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.
Meta launches Muse Wonder and Muse Wonder Full for sustained agent work
Meta released Muse Wonder through its model API and Muse Code, and Muse Wonder Full to Muse Code subscribers. The models emphasize persistent context, recovery after blockers and extended coding tasks. Meta reports strong agent evaluations; wider consumer rollout and safety-report additions were still forthcoming in the launch announcement.
- Full item contribution
- +0.11
- Muse Wonder equal share
- +0.05
Model score history
-
R1
Doom Score 84.9
Backfilled the exact October 12 release from the dated launch and model card, absent from the fresh catalogue.
30 Oct 2026