Muse Spark 1.3
Muse Spark 1.3 is Meta's proprietary hosted model for the Muse personal agent. It is trained for zero-shot tool use, long-context and long-trajectory instruction following, prompt-injection awareness, multi-agent coordination, browser operation, tool creation, and self-editing.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why this model scores 85.3
Meta describes a highly capable agent model that works unattended, launches subagents, builds tools, edits itself, uses a browser, and acts through connected services. Broad consumer rollout makes deployment high, and its access to private data and external actions creates substantial misuse exposure. Control difficulty is moderated by a separately isolated runtime, credential separation, Sentinel-controlled egress, scoped approvals, adversarial testing, and bug bounties.
News tied to Muse Spark 1.3
The model score of 85.3 rates this model's risk profile. The overall Doom Index of 67.9 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 personal agent for email, purchases, and long-running tasks
Meta launched Muse in the US as a consumer agent that can send email, book travel, fill forms, negotiate, make purchases, build tools, coordinate subagents, and continue working after the user closes the app. A separate Sentinel controls network access and sensitive actions.
- Full item contribution
- +0.14
- Muse Spark 1.3 equal share
- +0.14
Model score history
-
R1
Doom Score 85.3
Adds the exact Muse Spark 1.3 model used by Meta's newly released personal agent and its documented control architecture.
10 Sept 2026