Muse Wonder is available through Meta's hosted API and Muse Code. Wonder Full is a separate model on higher Muse Code plans; it is not a public pay-as-you-go API. API and coding subscriptions also have different data-use defaults.
RELATED BUYER PROFILEWonder API price and access evidencePricing · access · strengths · limitations →CONTINUE THE DECISIONWonder Full subscription boundaryEvidence · workflow · next action →CONTINUE THE DECISIONMuse Code plans and data settingsEvidence · workflow · next action →CONTINUE THE DECISIONCompare API routes and evidenceEvidence · workflow · next action →CONTINUE THE DECISIONBuild a same-task pilotEvidence · workflow · next action →CONTINUE THE DECISIONBudget the complete taskEvidence · workflow · next action →What you need to know
- Meta announced Wonder and Wonder Full on 12 October
- Wonder has a hosted preview API; Full requires an upper Muse Code subscription
- API Standard and Muse Code's default Contributor mode differ
What is available now
Meta's announcement makes Wonder available through the Meta Model API and Muse Code, with Full in Muse Code for eligible subscribers. The consumer Muse app rollout is still described as forthcoming. ActiveMemory and ActiveFocus are supplier descriptions of context and sustained-work behaviour; HubAI has not measured their reliability. This is a hosted release, not evidence of downloadable open weights.
Wonder versus Full: choose the route first
The model guide lists muse-wonder-1.0 for the public API. muse-wonder-full-1.0 is a separate model, available in Muse Code Power Usage and Maximum, without a public metered API. Setting Wonder's reasoning to max does not select Full. HubAI analysis: a team that needs API integration should evaluate Wonder rather than assume it can call the subscription model from an existing backend.
API price and preview limits
The published Wonder rates are USD $4 per million input tokens, $0.20 cached input and $20 output. Search grounding adds $2.50 per 1,000 queries. The overview describes a preview API at https://api.meta.ai/v1 with a 1,048,576-token context. These are component prices and specifications, not a guaranteed task cost. UK provisioning, residency and VAT are not verified by the launch.
Muse Code plans are a different purchase
The coding application lists monthly USD prices of $5 Everyday, $15 High Usage, $50 Power Usage and $200 Maximum. Full is limited to the upper two plans. Check the current allowance and what happens when it is exhausted before comparing a subscription with metered API work. HubAI analysis: choose the smallest eligible plan that can complete a representative task; do not buy the largest allowance solely because the model is new.
Review the data-use switch before sharing code
Meta says Standard API content is not used for training; Wonder API does not offer Contributor mode. Muse Code subscriptions instead default to Contributor mode, which can be turned off in settings. Record the setting before connecting a private repository. Neither mode alone establishes a UK residency promise, zero retention or suitability for confidential client material. Review the applicable contract and organisation policy.
HubAI buyer verdict
Keep the comparison practical: give Wonder and the current tool the same small repository change, acceptance tests, permissions and time budget. If Full is relevant, repeat inside its subscription route. Count accepted changes, regressions, tool failures, retries, review minutes and billed usage. A remembered instruction is useful only if it remains correct and can be changed or removed. No independent HubAI Score is assigned; real visitors can rate their experience separately.
Evidence and limitations
Independent editorial coverage; not sponsored. Official sources were checked on 14 October 2026. The access matrix is a documentary comparison, not a completed benchmark. UK account access and sandbox behaviour have not been independently tested. The original illustration depicts the split between API and subscription access; it contains no performance ranking or traffic claim.
HUBAI VIEWChoose the access and data-use route before comparing capability. A model upgrade and a coding subscription are different purchases.
Buyer decision signal: 12 October launch · API preview and subscription boundaries
What to verify next
1Confirm Wonder or Full and the actual account route
2Review Contributor mode before sharing private code
3Confirm UK access, retention and processing region
4Check allowance exhaustion, tools and tax
5Run the same task and acceptance tests on both sides
6Keep a human reviewer and a reversible branch
Read the evidence
Capabilities, availability and prices can change. HubAI keeps analysis separate from the underlying official material.
01Meta: Muse Wonder launch — 12 October 2026Open source ↗02Meta: model identities and access routesOpen source ↗03Meta: API pricing and data-use modesOpen source ↗04Meta: Muse Code plans and official demonstrationOpen source ↗05Meta: API preview and implementation guideOpen source ↗
Models
Models
Models
Models