same here on team aib-ois, been like this since the 26th. maybe 1 in 15 calls gets thru, the rest come back 429 high demand with providerAttemptCount 0 so it never even touches the provider. our rate is tiny, about 1 call a minute. bought credits today thinking the free tier was the issue, nope, paid tier does the exact same thing. the support bot said the gateway health check has typesafe marked unhealthy so it just fast fails everything, but the endpoints api shows 100 percent uptime so somthing doesnt add up. hobby plan cant open a case so posting here instead. one of todays rejected ids if it helps, gen_01M3JF68E219411054WMFD52GH. is this per team or is everyone cooked rn

Same here. Team on Hobby with paid AI Gateway credits (bought 28 Sep 2026), typesafe-ai/jev via /typesafe/v1/systemone. I measured exactly one accepted call roughly every 300 s; every call in between returns 429 "upstream provider is currently experiencing high demand", regardless of payload size. It has been consistent all day (Sydney time). It would be great to hear from Vercel whether this is per-team throttling, what the limit is, and whether Pro changes it.

We are seeing the same issue on a Hobby team with purchased AI Gateway credits. Other models work with the same key, but a minimal typesafe-ai/jev request returns HTTP 429 with totalProviderAttemptCount: 0. Pinning providerOptions.gateway.only to ["typesafe-ai"] also returned 429.

Minimal reproduction: POST https://ai-gateway.vercel.sh/v1/evaluate

{"model":"typesafe-ai/jev","state":"The circle is red.","questions":{"red":{"type":"boolean","instructions":"Is the circle red?"}}}

Error: "The upstream provider is currently experiencing high demand. Please retry shortly."

Vercel's support assistant prepared a case, but the case form says Hobby accounts must use the community for AI Gateway support, so we are adding this report here.

Could Vercel staff clarify whether this is a known Jev routing or provider-health issue, a paid-tier activation issue, or a documented per-team restriction? What action is needed to restore access? We can provide a generation ID privately to staff for investigation.

Same issue. statusCode: 429, name: 'RateLimitExceededError', message: 'No access to this model at this time.'

Currently, the rate limit restricts us to only one request every 3–5 minutes, which has forced our work to halt. Please confirm whether this is an AI Gateway issue or a policy change from the provider, and provide a solution as soon as possible.

Same here too. Team on Hobby with paid AI Gateway credits. rate limits are killing the needs. Brought the credits 29-09-2026. unable to reach support due the hobby tier. its frustrating

We can reproduce the same issue on a Hobby team with purchased AI Gateway credits, even though the authenticated diagnostics show the account and model as available.

• GET /v1/credits confirms a positive balance.
• GET /v1/models?include_availability reports request_context_availability.http.status=available and typesafe-ai/jev model_eligibility.status=eligible.
• The unchanged official one-Boolean example and a choice+Boolean request both reproduce the issue.
• One request succeeded 313.9 seconds after the preceding success; repeating it 0.85 seconds later failed. A later pair again succeeded once and failed 0.45 seconds later.
• The failures return 429 rate_limit_exceeded, No access to this model at this time., and totalProviderAttemptCount=0. Pinning only typesafe-ai does not resolve it.

Latest failure generation: gen_01M3QKZRV1MCAJ83JYNSQ5EY2N at 2026-09-29 22:18:56 UTC.

The Vercel Support Agent directed us here and said it could not inspect or resync the internal Gateway controls. Could the AI Gateway team check the admission decision for this generation and confirm whether this is a Jev-specific interval, a reliability safeguard, or paid-tier entitlement propagation? What remedy is available without an additional purchase? We can provide team details privately to staff.

Hey all,

Access to Typesafe-ai/jev may be temporarily limited for some accounts as part of safeguards that protect service reliability. When this happens, requests return a 429 response.

Please try again later or use another available Model in the meantime. Once your account passes verification checks, the rate limit should be lifted automatically.

We appreciate your understanding.