Post-Cutoff
Search For AI agents

MetaMuse familyPreviewReasoning modelReleased 12 Oct 2026

Muse Wonder Full

The larger of Meta’s two frontier models released Oct 12, 2026. There is no API access; use it through Muse Code.

Model idmuse-wonder-full-1.0

PricePrice not published

Checked12 October 2026

No published training cutoff

At least0events since its release on 12 October 2026

0 historic, 0 major. 0 days of news.

Meta doesn’t publish a training cutoff, so we count from its release date. The real gap is larger.

How to call it

Where to use Muse Wonder Full, with its model id.

Providers, model ids, endpoints and docs for Muse Wonder Full
ProviderModel idEndpointDocs
Muse Code (Power Usage and Maximum Usage plans only) muse-wonder-full-1.0 Docs
Model page

Pricing

Price not published.

No price source linked. Checked 12 October 2026.

Specs

Input
Text, images, video, audio and PDF
Output
Text
Context window
1,048,576 tokens
Max output
Not published
Open weights
No
Licence
Proprietary
Training cutoff
Not published
Released
12 October 2026
Status
Preview
Type
Reasoning model

What stands out

What its maker marketed at launch, and what people found later, each with its source.

  • Long-horizon coding

    DeepSWE 1.1 79.0 (Datacurve) and SWE-Marathon 57.5 at xhigh (Abundant AI), the top scores in Meta’s table; Artificial Analysis Coding Agents Index 68 in Muse Code, level with Claude Code with Sonnet 5.5.

    Source: research.meta.ai

  • Legal, finance and business-workflow agents

    Harvey LAB-AA 13.1, Vals Finance Agent V2 59.1 and AutomationBench-AA 45.4, each ahead of Claude Fable 5.1, GPT-6 Astra and Claude Opus 5.5 in Meta’s table.

    Source: research.meta.ai

  • Active Memory and Active Focus

    Cross-conversation memory (MemoryBench 85.5, internal) and persistence on blocked tasks, shared with Muse Wonder.

    Source: research.meta.ai

Notes

Not on the Meta Model API and no per-token price; available only with a Muse Code subscription ($50 Power Usage, $200 Maximum Usage per CNBC and Engadget). A separate model, not a reasoning level of Muse Wonder. Artificial Analysis Intelligence Index v4.3: 56 at max (Opus 5.5: 58; Fable 5.1 and GPT-6 Astra: 53); Vals Index #5. Model card: unmitigated model may reach Meta’s ‘high risk’ threshold for chemical and biological misuse; mitigations applied. Knowledge cutoff and max output not published.

Sources: https://dev.meta.ai/docs/models , https://dev.meta.ai/models/muse-wonder/ , https://research.meta.ai/static/muse-wonder-model-card , https://x.com/artificialanlys/status/2109697083519447118