⌘K
8008makoto 9443 I am seeing the same error on my Ortus Lab Hobby team as of 4 October 2026. A single connection-only request to POST /v1/evaluate, using typesafe-ai/jev with providerOptions.gateway.only = ["typesafe-ai"], returned HTTP 403 / no_providers_available. No retries were made. Gateway generation: gen_01M42Z5VSGVE2NZHP2PFH7B9JS Request ID: kix1::cle1::xxhnt-1791101169365-b47ac4afa936 Timestamp: 2026-10-04 08:06:09.456 UTC The authenticated dashboard shows provider allowlist, model allowlist and ZDR disabled. The request log shows TypeSafe AI / Jev and no reported usage. Provider reachability and the
Gabi Norrman Hi Vercel team, I'm experiencing the same issue with typesafe-ai/jev-1.14.0 on my Hobby team. POST /v1/evaluate returns HTTP 403. The AI Gateway key is recognised. The model page says the TypeSafe AI provider is unavailable to my team. Zero Data Retention is disabled. The request reaches TypeSafe AI but reports no token usage. AI Gateway Logs confirm $0 charged. Observed on 3 November 2026. Could you confirm whether this is a provider availability, account entitlement or provisioning issue, and whether access can be restored on Hobby? I can provide the Gateway request ID privately if needed. T
Anshuman Bhardwaj Hi everyone, as per our pricing page (vercel.com/ai-gateway/models/jev), you cannot use free credits for Jev models on either TypeSafe or DigitalOcean providers. Let me know if you have another question.
Amy Egan Hey @paulmain79. Do you have a support case open? The community isn't able to issue refunds. That kind of thing is limited to the support team. If you don't have a case yet, please open one at vercel.com/help so the support team can assist!
Paulmain79 Hi Amy, thank you so much for coming back to me, and sorry for the slow reply. I've now opened a case: #CSE_01iu4vykeYDA9JVJ (Billing > Refund Request, Paul Main's projects, Pro). The case has the full picture: how the Preview build costs grew from May to October as our automated coding agents scaled up, the October Build CPU figure, and the three changes we've made (an ignoreCommand, on-demand Previews and Spend Management). I've had a lot of automated replies while trying to get this looked at, so I'd be really grateful if a member of the Billing team could take a look personally. We're a sm
Amy Egan Hey @jennie369. It looks like you've been working with a support engineer on your case. Please continue to follow up with them to get this solved 🙏
Jenniec Hi Amy, thanks for checking in. I have been working with the support engineer, but the case isn’t resolved. They closed it with a 25% goodwill refund ($134.64) and said that’s the maximum. $489.52 is still unrefunded. The issue is consent. My projects were explicitly set to fixed Basic/Standard machines, and the system overwrote that choice the minute I upgraded (isSystemInitiated: true, “because your team’s plan changed”). The upgrade screen said nothing about changing build machines or adding on-demand Build CPU charges. On Oct 3 it also auto-upgraded yinyang-web to 8 CPUs without asking. At
Anshuman Bhardwaj Hi @luke-9487, I'm sorry that you had to face this issue. Since it's related to billing, I'd recommend opening a billing-related case at vercel.com/help and explaining everything there. Our support team is better equipped with handling billing issues.
Anthony Shew Copy-pasting from community.vercel.com/t/introducing-vercel-ci/51548#post_5 Hey, Luke, sorry for the trouble. We had a bug that ended up overbilling some folks. We've applied credits to the affected customers so you'll see our mistake covered on your next invoice. Sorry again for the trouble! Additionally, I see you're also mentioning here that you had tasks hanging open on this post. That was the same bug, and is now remediated.
