Where can I use Muse Wonder today?

Meta is offering Muse Wonder through Meta Model API and Muse Code, and Muse Wonder Full through Muse Code for subscribers. General users will access Muse Wonder through the Muse app.

What are Active Memory and Active Focus?

Active Memory holds important user information across conversations and recalls related details on its own. Active Focus keeps working toward a goal on difficult tasks, trying other approaches and recovering from errors.

How does Muse Wonder Full compare with rival models?

On Harvey LAB-AA, Muse Wonder Full scored 13.1, beating GPT-6 Astra's 8.6 and Claude Opus 5.5's 4.2. On DeepSWE 1.1 it scored 79.0, above Claude Opus 5.5's 74.2 and GPT-6 Astra's 74.1.

Large Language ModelsITmedia AI＋3m agoOct 13, 2026, 10:01 JST

Meta launches Muse Wonder and Muse Wonder Full via Meta Model API

Send on LINE

3 Key Points

What happened
Meta unveiled Muse Wonder and the higher-tier Muse Wonder Full on October 12, calling them its largest and most capable models, and started offering them the same day via Meta Model API and Muse Code.
Why it matters
Meta is positioning the models as its most capable yet for agent tasks that act on a user's behalf, and it will also bring them to its personal AI agent Muse, starting with subscribers.

WHO IT HITSDevelopers using Meta Model API and Muse Code gain access to the new models immediately. Meta's Muse subscription users will get them next, though Meta has not said when or for which plans.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

Meta's new Muse Wonder series builds on the foundation laid by Muse Spark, announced in April, by scaling up both pre-training and reinforcement learning, and it was trained at the gigawatt-scale data center Prometheus. Meta says that because the base model is larger and more powerful, Muse Wonder performs better than Muse Spark 1.3's tendencies would predict even on unfamiliar tasks that were not a focus during training.

The launch lands in an already crowded field. Meta's own model card compares Muse Wonder against OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 and Claude Opus 5.5, and the results are mixed: Meta's models lead on legal-work and long-horizon coding benchmarks, while Claude Opus 5.5 scores highest on health questions, terminal operation, building programs from scratch, and Artificial Analysis's overall index.

Safety work followed Meta's Superintelligence Scaling Framework. Cybersecurity and loss-of-control areas were rated at most moderate risk before mitigation, but because chemical and biological risks could not be ruled out as high, Meta added further training and system-level classifiers to bring all areas to moderate or below.

FAQ

Where can I use Muse Wonder today?: Meta is offering Muse Wonder through Meta Model API and Muse Code, and Muse Wonder Full through Muse Code for subscribers. General users will access Muse Wonder through the Muse app.
What are Active Memory and Active Focus?: Active Memory holds important user information across conversations and recalls related details on its own. Active Focus keeps working toward a goal on difficult tasks, trying other approaches and recovering from errors.
How does Muse Wonder Full compare with rival models?: On Harvey LAB-AA, Muse Wonder Full scored 13.1, beating GPT-6 Astra's 8.6 and Claude Opus 5.5's 4.2. On DeepSWE 1.1 it scored 79.0, above Claude Opus 5.5's 74.2 and GPT-6 Astra's 74.1.

ITmedia AI＋Read Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.