Troubleshooting · Cursor

Cursor: Unable to Reach the Model Provider — Causes and Fixes

Cursor reports it cannot reach the model provider when the request dies before an HTTP response: wrong base URL, network path, or a provider outage. How to tell which one you have.

What you see

We're having trouble connecting to the model provider. This might be temporary - please try again in a moment.

Why it happens

  • The custom base URL is wrong at the transport level — a typo in the host, a scheme mistake, or a URL that resolves nowhere — so no HTTP response ever arrives.
  • The network path is blocked: a proxy, VPN or firewall between Cursor and the endpoint drops the connection.
  • The provider itself is briefly down. In that case nothing on your side changed and the error clears on its own.
  • An override URL was pasted with a trailing path segment the provider does not serve, so the TLS or HTTP layer fails before any API error can be returned.

How to fix it

  • Test the exact base URL outside Cursor with curl. If curl cannot connect either, the problem is the URL or the network, not Cursor.
  • Set the OpenAI base URL override to https://router.apitoken.sale/v1 exactly and let Cursor append the API path itself.
  • If curl works but Cursor does not, check whether Cursor is routed through a proxy or VPN the terminal is not using.
  • If nothing changed on your side and the error is new, wait a few minutes — transient provider incidents produce exactly this message.

Prove the endpoint is reachable

curl https://router.apitoken.sale/v1/models \
  -H "Authorization: Bearer sk-pool-•••"

Related searches

  • cursor unable to reach model provider

FAQ

Is "unable to reach model provider" the same as a 401 or 429?+
No. This message means no HTTP response arrived at all. A 401 or 429 means the provider answered and rejected the request — different layer, different fix.
Cursor worked yesterday and fails today with no changes — what now?+
That pattern is a transient outage or a network-path change (VPN, proxy, captive portal). Verify with curl first; do not rewrite a configuration that already worked.
What base URL should the OpenAI override use?+
For this gateway, https://router.apitoken.sale/v1 — with /v1 and nothing after it. Cursor adds the API path itself, and an extra path segment fails.

Skip the broken setup

apiToken.sale serves GPT and Gemini models through OpenAI-compatible, Messages (Anthropic format) and native Gemini endpoints — standard SDKs, one key, one prepaid balance. Point your tool's base URL at it, and the config on this page works as written.