Uploaded October 2026 | Updated November 2026, 3 days ago
Meta's Muse Wonder scores higher on agent tasks. Its memory claim rests on a Meta-internal test.
Meta launched Muse Wonder and Muse Wonder Full on October 12, 2026. Wonder is available through Meta's Model API and on all Muse Code plans. Full is only on the Power Usage and Maximum Usage plans. Both models list a 1,000,000-token context window.
Artificial Analysis, an independent evaluator, reports Intelligence Index v4.3 scores at max effort: Muse Spark 1.3 at 48, Wonder at 52, Wonder Full at 56. Its agent evaluations also show substantial gains. These are the evaluator's own results; they do not establish reliable memory in production or superiority on every task.
Meta describes Active Memory as retrieving relevant context across conversations, and Active Focus as persisting through blockers and checking before important actions. Those are vendor descriptions. Meta's MemoryBench is an internal test using conversations of 64,000 to 100,000 tokens, including distractors and corrections. That does not show whether memory holds up across months of real work.
The Muse personal-agent rollout was still forthcoming as of October 13, 2026.
My take (opinion): test whether it remembers a corrected requirement, recovers from a blocker, and asks before taking a consequential action.
What would you check first before trusting an AI agent with months of work?
Event: October 12, 2026. Research checked: October 13, 2026.
Byte’s Report — research with Byte, commentary by Dave.
Sources:
Introducing Muse Wonder and Muse Wonder Full: Frontier Models for Agents — Meta — 2026-10-12
about.fb.com/news/2026/10/introducing-muse-wonder-models
Muse Wonder: A Frontier Model For Personal & Professional AI Agents — Meta Research — 2026-10-12
research.meta.ai/blog/muse-wonder
Models — Meta Model API — undated; checked 2026-10-13
dev.meta.ai/docs/models
Muse Wonder Model Card — Meta Research — undated; checked 2026-10-13
research.meta.ai/static/muse-wonder-model-card
Muse Wonder and Muse Wonder Full: a material jump in intelligence over previous Meta models — Artificial Analysis — 2026-10-12
artificialanalysis.ai/articles/muse-wonder
#Meta #MuseWonder #AIAgents #BytesReport
Meta's Muse Wonder scores higher on agent tasks. Its memory claim rests on a Meta-internal test.
Meta launched Muse Wonder and Muse Wonder Full on October 12, 2026. Wonder is available through Meta's Model API and on all Muse Code plans. Full is only on the Power Usage and Maximum Usage plans. Both models list a 1,000,000-token context window.
Artificial Analysis, an independent evaluator, reports Intelligence Index v4.3 scores at max effort: Muse Spark 1.3 at 48, Wonder at 52, Wonder Full at 56. Its agent evaluations also show substantial gains. These are the evaluator's own results; they do not establish reliable memory in production or superiority on every task.
Meta describes Active Memory as retrieving relevant context across conversations, and Active Focus as persisting through blockers and checking before important actions. Those are vendor descriptions. Meta's MemoryBench is an internal test using conversations of 64,000 to 100,000 tokens, including distractors and corrections. That does not show whether memory holds up across months of real work.
The Muse personal-agent rollout was still forthcoming as of October 13, 2026.
My take (opinion): test whether it remembers a corrected requirement, recovers from a blocker, and asks before taking a consequential action.
What would you check first before trusting an AI agent with months of work?
Event: October 12, 2026. Research checked: October 13, 2026.
Byte’s Report — research with Byte, commentary by Dave.
Sources:
Introducing Muse Wonder and Muse Wonder Full: Frontier Models for Agents — Meta — 2026-10-12
about.fb.com/news/2026/10/introducing-muse-wonder-models
Muse Wonder: A Frontier Model For Personal & Professional AI Agents — Meta Research — 2026-10-12
research.meta.ai/blog/muse-wonder
Models — Meta Model API — undated; checked 2026-10-13
dev.meta.ai/docs/models
Muse Wonder Model Card — Meta Research — undated; checked 2026-10-13
research.meta.ai/static/muse-wonder-model-card
Muse Wonder and Muse Wonder Full: a material jump in intelligence over previous Meta models — Artificial Analysis — 2026-10-12
artificialanalysis.ai/articles/muse-wonder
#Meta #MuseWonder #AIAgents #BytesReport










