Observed facts: Cursor’s documentation lists Muse Wonder as model ID muse-wonder-1.0, with Meta as provider, a 300,000-token context window, and up to 1 million tokens in Max Mode. Observed facts: The page says it is not included in Cursor plan usage, lists $4 per million input tokens and $20 per million output tokens, and notes that some Teams and Enterprise administrators must enable Meta access.
Mechanism: Cursor exposes the model through its existing agent environment, where the documentation says it can use file search, editing, shell, browser, and other agent tools. The practical unit is therefore not just a model endpoint: users can apply it within a coding harness, but the provider setting and metered model charges remain separate constraints. Why now: The model’s appearance in Cursor’s catalog turns Meta’s coding model into an option within a workflow builders already use, making the access and cost conditions immediately relevant to model selection.
First implication: Teams can evaluate Muse Wonder against their coding tasks inside Cursor, but should account for usage outside included plan allowances and verify provider access before rollout. Evidence boundary: These details come from Cursor’s product documentation; the page does not independently establish benchmark superiority, user adoption, or performance in production workloads.
First watch condition: Watch for changes to Cursor’s model listing, provider-enablement requirements, or stated token rates, and for independently reproducible task-level comparisons. This begins a distinct analytical record because no prior New Runtime record was supplied; it records a specific model’s availability and operating conditions in Cursor, not a general claim about the broader Muse product family.