Is Cursor down?
The live status from status.cursor.com is below. During a model outage Cursor shows "Unable to reach the model provider" or "High Load". Switch to Auto or another model in the model picker, start a new chat, or run Settings > Network > Run Diagnostics if the status is green.
Live Cursor status
Checked
All systems operational
- IDEOperational
- CLIOperational
- Cloud AgentsOperational
- cursor.comOperational
- Last incident: Investigating service degradation — Grok Botresolved Oct 14
Source: status.cursor.com. Read on our server about every 2 minutes.
Outage, or your setup?
| What Cursor shows | What it means |
|---|---|
| "Unable to reach the model provider" | "We're having trouble connecting to the model provider. This might be temporary - please try again in a moment." A failure between Cursor and the model's provider. Retry; if it repeats, switch model or start a new chat, which has fixed it in some cases |
| "High Load" | "We're experiencing high demand for the selected model right now. Please switch to Auto, another model, or try again in a few moments." That model is out of capacity (ERROR_RESOURCE_EXHAUSTED). Switch model |
| "Connection failed" | "If the problem persists, please check your internet connection or VPN." Usually your network, proxy or HTTP/2. Run Settings > Network > Run Diagnostics |
| A usage limit notification | Your plan's included usage, not an outage. See Cursor usage limits |
Five minutes into a Cursor outage
- Check the IDE, CLI and Cloud Agents rows above, and the incident title: Cursor names the model when one model is affected.
- If one model is failing, open the model picker in the chat or agent panel and choose another one, or press Cmd + / to cycle models. Auto lets Cursor pick.
- If the status is green and only you fail, start a new chat, then run Settings > Network > Run Diagnostics and try with HTTP/2 off.
- If Cursor's own services are down, the editor still edits files; a terminal agent with its own provider keeps the agent work going.
Cursor with your own API key
| What it covers | |
|---|---|
| Where | Settings > Models: OpenAI, Anthropic, Google, Azure OpenAI or AWS Bedrock keys |
| What runs on it | Chat and agent with those models. Tab completion keeps using Cursor's own models |
| Plans | Own keys are not available on every plan; Plus does not include them |
| The catch | Requests still pass through Cursor's servers, so an outage of Cursor's own backend stops them too |
From cursor.com/help on own API keys, October 15, 2026. With Override OpenAI Base URL, a gateway such as Standard Compute can serve chat and agent requests; see Cursor with your own API key.
What Standard Compute does when a provider fails
- When Cursor sends a request to Standard Compute, it goes through OpenRouter to a provider order we review per model. If that provider fails, OpenRouter can send the request to another provider of the same model.
- Each failed call is retried twice on a 429, a 5xx or a network error, after 0.5 and then 1 second, or after the provider's Retry-After up to 8 seconds. If every attempt fails, your agent gets HTTP 503 with "The model is temporarily unavailable. Please retry shortly."
- A specific model ID always stays that model: failover changes the provider, never the model.
The catch
Own keys and gateways help when a model provider fails, not when Cursor's own backend does: requests still go through Cursor's servers. And when a model's maker is down at the model level, every provider of that model may be affected; switch to a model from another vendor.
Where Standard Compute fits
Standard Compute is built to make coding agents cost less: its router sends each step to the cheapest model that can do it, on providers we vet for price, uptime and speed, and a failed provider falls back automatically to another provider of the same model. An opt-in Compliance rule keeps a key, a team or the whole company on endpoints in Europe, the United States or both, and we don't train on your prompts or store them. One key at https://api.stdcmpt.com/v1 serves Claude, GPT, Gemini, DeepSeek, GLM and Kimi models, from $19 a month (USD).
FAQ
Is Cursor down or is it just me?
Check status.cursor.com for the IDE, CLI and Cloud Agents rows. If they are green, start a new chat, then run Settings > Network > Run Diagnostics and try with HTTP/2 off; "Connection failed" is most often the network, a proxy or a VPN.
What does 'Unable to reach the model provider' mean in Cursor?
Cursor could not get an answer from the model's provider. It is often temporary: retry, switch model or start a new chat. If many users see it at once, an incident appears on status.cursor.com.
What does Cursor's 'High Load' error mean?
The selected model has no spare capacity right now. Switch to Auto or another model, or try again in a few moments. It is not a usage limit on your account.
Does my own API key keep Cursor working during an outage?
Only when the outage is at a model provider. Requests with your own key still pass through Cursor's servers, so an outage of Cursor itself stops them too, and Tab completion always uses Cursor's own models.