Meta announces Muse Wonder and Muse Wonder Full with Active Memory and Active Focus
Meta unveiled two frontier‑class AI models, Muse Wonder and Muse Wonder Full, on October 17, 2026. Both are built for agent functions and will be accessed through the Meta Model API. They introduce Active Memory, which can retain conversation and work history for days, weeks, or months, and Active Focus, which drives the model to try alternative approaches and keep working when obstacles appear.
What you can do with it
AI at Work →Draft reports with Meta Model API
What you get You can let the AI handle multi-step projects without constant prompts.
AI agents retain long-term context and persist through obstacles.
- Who for
- marketers
- Cost
- Price not stated
- Effort
- minutes
Use it for
- Draft reports
- Manage customer follow-ups
- Run data analyses
How to set it up · From the article
- Monitor Meta Model API updates for pricing and task optimisation
- Identify tasks to delegate to agent‑type AI such as long‑term follow‑ups
- Redesign workflows to increase hands‑off points
Watch out Only available through Meta Model API, pricing not disclosed
Meta Model API in the tool directory · 2 changes covered
Key points from the news
- Meta unveiled Muse Wonder and Muse Wonder Full, agent‑focused models with Active Memory and Active Focus.
- Muse Wonder Full matched Claude Opus 5.5 and GPT‑6 Astra on some benchmark scores, according to Meta.
- Features aim to boost AI autonomy, complex‑task completion, and market competition, potentially lowering costs for users.
Meta says the features will raise AI autonomy, letting the system act like a long‑term business partner, and improve completion of complex, multi‑step tasks. The company claims Muse Wonder Full achieved benchmark scores comparable to Anthropic’s Claude Opus 5.5 and OpenAI’s GPT‑6 Astra. The announcement also suggests tighter competition could bring higher‑performance, lower‑cost services.
The author recommends three immediate actions for businesses: follow the Meta Model API for pricing and task‑optimisation details, list tasks that could be delegated to an agent‑type AI such as long‑term customer follow‑ups, and redesign workflows to increase hands‑off points. Readers are invited to contact the author via LINE for personalized strategy.
Model pages: Muse Wonder → · Claude Opus 5.5 → · GPT-6-Astra →
The story so far
4 episodes →- Meta announces Muse Wonder and Muse Wonder Full with Active Memory and Active Focusthis story
[Breaking News] For AI Users! 3 Impacts of Meta's New AI
note.com · 17 October 2026
Loading the full article…
This text was published by note.com and written by モモン先生 AIと融合しているAIコンサル. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Modders report DLSS 5 yields up to 90% performance boost on unsupported games · 2 src
- DeepSeek releases DeepSeek-V4.1-Flash-1017, scores 77.0 on DeepSWE v1.1 and opens Search and Fetch APIs · 17 src
- Google makes Gemini 4 Argon generally available to AI Pro and Ultra subscribers · 38 src
- OpenAI makes Prompt-to-Voice generally available for developers · 2 src
- NVIDIA releases Nemotron 3.5 Super VL and Bolt edge models for enterprise agents · 12 src
Comments
via GitHub Discussions