What's happening About 98% of my requests to typesafe-ai/jev via AI Gateway return 429 "No access to this model at this time" (type: rate_limit_exceeded, providerAttemptCount: 0), so they're refused before reaching any provider. This happens even at ~10 requests/minute. The few requests that get through work normally (~0.5 s each).

What I've tried

• Added a card and bought $10 of AI Gateway credits (29 Sep, ~17:58 BST). Support confirmed paid credits should remove the Gateway's free-tier rate limits, with up to 15 minutes to propagate.
• Waited well over 15 minutes, then created a new API key under the same team that shows the credit balance. Batches of 20 requests at 18:29 and 18:34 BST got 1, 0 and 0 successes.
• Not specifying any provider (forcing typesafe-ai made no difference).
• The team's Requests log shows these as 429s with $0.0000 cost (screenshot attached).

Setup: Hobby plan with paid AI Gateway credits, AI SDK ai@7.0.119 with experimental_evaluate, calling from a local Node script.

Generation IDs: gen_01M3Q28C6NAEXJ6GW564NDQTGV, gen_01M3Q276D35PFTP45VT1VZ9C5H, gen_01M3Q1KSJ9VTQH71MZT9NFE06T

Is the paid tier not being applied to my team, or is there a separate limit on Jev? Is anyone else seeing this?