General editorial illustration for Agentic AI.

The briefingAgentic AI

Topic illustration

Meta releases Muse Wonder and Muse Wonder Full agent models

By Ashlr.AIAI-assisted briefingPublished Oct 13, 2026, 10:15 AM UTC2 min read

Meta introduced Muse Wonder and Muse Wonder Full, its latest models for personal and professional agent tasks. The models include Active Memory for cross-conversation context and Active Focus for persistent goal pursuit with error recovery. Availability is through the Meta Model API and Muse Code now, with broader Muse rollout planned for subscribers.

What changed

Meta released Muse Wonder and Muse Wonder Full as its most capable models for agentic tasks in domains including law, finance, health, and coding. The models add Active Memory to retain task-relevant details across conversations and Active Focus to maintain progress on difficult tasks, attempt alternatives, recover from errors, and seek clarification.

Browse all stories

AI-generated conceptual editorial illustration. Not a documentary photograph.

The briefingAgentic AI

GitHub adds computer-use preview to Copilot CLI and desktop app

Oct 3, 20262 min read

GitHub released a public preview of computer use that lets Copilot agents read, click, type, scroll, and drag inside macOS and Windows applications, including legacy GUI software without APIs.

Access is available today via the Meta Model API and Muse Code. The models will power additional Muse features, starting with subscribers. A full safety report aligned with the Meta Superintelligence Scaling Framework is forthcoming.

Safety training targets prevention of unauthorized private information sharing, refusal of harmful requests, and support for defensive security work such as vulnerability detection. Developers retain the ability to use the models for code analysis and system hardening.

Meta states the models were built on prior Muse Spark groundwork and deliver frontier performance on multistep real-world workflows that unfold over time.

What it means for your work

For builders

Builders evaluating agent frameworks can test Muse Wonder through the Meta Model API to assess Active Memory retention and Active Focus persistence on domain tasks such as lease analysis or application development. An illustrative scenario is directing an agent to reconcile insurance claim documents across multiple sessions while verifying it surfaces only permitted details and requests clarification on ambiguous policy language. Unknown factors include exact token pricing, rate limits by plan, and measured latency on long-running agent loops compared with prior Muse Spark deployments.

For business

Operators reviewing agent deployments should confirm subscription eligibility and data-handling terms before routing messages, calendars, or documents through Muse-powered features. An illustrative scenario is piloting the models on internal finance workflows that span several days, recording which conversation elements are retained in Active Memory and confirming refusal behavior on requests involving restricted health data. Material unknowns are retention periods for Active Memory contents, regional availability restrictions, and the scope of the upcoming safety evaluation report.

Put it to work

Request Meta Model API access documentation and review the subscriber rollout schedule for Muse integration. Record the specific agent tasks tested, observed context retention behavior, and any clarification prompts issued by the model. Compare these observations against current agent tooling to determine whether the Active Memory and Active Focus capabilities justify migration or parallel evaluation.

Go to the source

Source material behind this briefing. Our practical implications are Ashlr’s interpretation.

Introducing Muse Wonder and Muse Wonder Full: Frontier Models for Agents (opens in a new tab)
Meta NewsroomOct 12, 2026
Meta releases Muse Wonder AI models built to run personal and professional agents (opens in a new tab)
SiliconANGLEOct 12, 2026

This briefing was prepared with AI assistance from the linked sources. The analysis may include interpretation; consult the original sources for complete context. Cover artwork is illustrative.

Editorial approach Suggest a correction

The briefingAgentic AI

Google's agent kit reaches 1.0 in Kotlin, with on-device pieces

Sep 20, 20263 min read

Google's Agent Development Kit reached 1.0 for Kotlin, adding Android and JVM targets. InfoQ reviewed it on September 20; Google's own post is dated September 9, and one on-device path is still beta.

The briefingAgentic AI

Google launches Gemini 3.8 Live voice models

Sep 18, 20263 min read

Google’s new voice models can keep a conversation moving during background tasks. The useful test is whether dialogue, tool outcomes and access requirements line up.

Meta releases Muse Wonder and Muse Wonder Full agent models

What changed

Keep exploring

GitHub adds computer-use preview to Copilot CLI and desktop app

What it means for your work

For builders

For business

Put it to work

Go to the source

Google's agent kit reaches 1.0 in Kotlin, with on-device pieces

Google launches Gemini 3.8 Live voice models