Muse Wonder
An AI model released by Meta Superintelligence Labs and part of the Muse series. It is positioned as a flagship or standard-tier model intended to be competitive with leading models in the industry.
Core Facts
Muse Wonder is offered through the Meta Model API and Muse Code; Muse Wonder Full is for Muse Code subscribers.
Muse Wonder offers six reasoning levels: minimal, low, medium, high, xhigh, and max.
The Muse Wonder series includes two model variants: Wonder (a streamlined version optimized for speed) and WonderFull (the full version provided via API).
Timeline (last 90 days)
Given the same reverse-engineering prompt and files, Meta's Muse Wonder model froze outright and failed to complete the port.
The lack of specific data (such as the store's URL, technical details or user feedback) makes it difficult to assess the store's actual impact.
Muse may be an AI creative platform, and Muse Wonder may be a derivative project focused on e-commerce or fashion, but this is only speculation.
The tweet does not provide details about the store's specific features, technical background or operating model.
A tweet claims that Muse and Muse Wonder have jointly built an online store where graphic T-shirts of global craftsmanship can be purchased.
Currently, independent sources of information about the store are limited, based mainly on a brief tweet, and multi-source cross-verification cannot be performed.
On the CursorBench benchmark, Muse Wonder scored 46.1% at default effort.
At present, details of Muse Wonder's architecture, its training data and broader benchmark results have not been disclosed.
Meta Superintelligence Labs' model Muse Wonder has officially landed on the Cursor platform.
The Muse series is Meta Superintelligence Labs' model family for developers and productivity scenarios, and Wonder is the member of that family aimed at coding assistance.
40 more timeline events
All Facts (20)
Muse Wonder is offered through the Meta Model API and Muse Code; Muse Wonder Full is for Muse Code subscribers.
80%VerifiedMuse Wonder offers six reasoning levels: minimal, low, medium, high, xhigh, and max.
70%VerifiedThe Muse Wonder series includes two model variants: Wonder (a streamlined version optimized for speed) and WonderFull (the full version provided via API).
65%VerifiedMeta released the frontier model Muse Wonder on October 12, and launched Wonder Full at the same event.
65%Partially VerifiedMuse Wonder supports a 1 million Token context window.
75%Partially VerifiedMuse Wonder offers three Meta API request formats (Responses, Chat Completions, Messages), Muse Code, and Vercel AI Gateway and Cloudflare Workers AI integrations.
70%Partially VerifiedMuse Wonder natively accepts text, image, video, and audio input, whereas Claude Opus 5.5 only supports text, image, and PDF.
65%Partially VerifiedMuse Wonder supports a 1,048,576 Token context window (per request).
65%Partially VerifiedCached input still counts toward TPM; caching is a billing discount, not a rate-limit discount.
65%Partially VerifiedMuse Wonder focuses on two major features: Active Memory and Active Focus.
65%UnverifiedClaude Code runs on top of Muse Wonder via the Messages endpoint.
60%UnverifiedFor developers familiar with the OpenAI SDK, migrating to Muse Wonder requires only replacing base_url and api_key.
60%UnverifiedClaude Opus 5.5 is the absolute flagship of Anthropic's product line, while Muse Wonder is positioned as the smaller of Meta's two models, with its larger version being Muse Wonder Full (available only via Muse Code).
60%UnverifiedMuse Wonder natively supports video and audio input, while Claude Opus 5.5 does not.
60%UnverifiedReasoning Tokens count toward max_output_tokens and billing, and output Tokens likewise consume TPM.
60%UnverifiedEach model has an independent quota pool, so high traffic on Muse Wonder will not crowd out Muse Spark's channel, and vice versa.
60%UnverifiedInstruction context injected by Meta does not count toward TPM and does not appear in the usage field.
60%UnverifiedMuse Wonder's rate limits are shared per team and per model, rather than calculated independently per API key; all keys within a team consume the same quota pool.
60%UnverifiedIn multi-agent concurrency scenarios, it is recommended to keep at least 40% of TPM headroom, because agent turns are bursty and a retry storm can double request volume within a minute.
60%UnverifiedMuse Wonder's token-based responses carry x-ratelimit-* headers; transcription responses do not carry these headers, whereas Wonder responses do.
60%