Seeing the same mismatch on the release build: macOS 27.0.1 (26A434), Xcode 27.0 (27A266a), entitled app (com.apple.developer.private-cloud-compute).
On 2026-10-16, after ~450 development requests, every PrivateCloudComputeLanguageModel request failed with PrivateCloudComputeLanguageModel.Error.quotaLimitReached. At the same moment a fresh PrivateCloudComputeLanguageModel().quotaUsage reported status = .belowLimit(isApproachingLimit: false), isLimitReached = false, resetDate = nil, availability = .available. isApproachingLimit never became true earlier in the day. Requests worked again the next morning. Filed as FB25222430 (references FB24515295).
Related: the Xcode scheme option "Simulated Apple Foundation Models Availability" set to "Quota Usage Limit Reached" or "Approaching Quota Usage Limit" has no effect on PrivateCloudComputeLanguageModel on macOS — quotaUsage stays belowLimit and requests are served normally. Filed as FB25222386 with a sysdiagnose.
For now our app treats the quotaLimitReached error itself as the signal: it records the limit from the error (with resetDate and limitIncreaseSuggestion when present) and falls back to SystemLanguageModel.
Topic:
Machine Learning & AI
SubTopic:
Foundation Models
Tags: