Error codes Models
PULSE-MODEL-003 The model server responded with an error
What does this mean
The request reached the model server, but the server was unable to fulfill it and returned an error. The agent's progress is stopped at this request; what the agent managed to do before him remains in the working copy of the task.
Why does this happen
Failure on the server side: the provider is overloaded or restarted, the local server (Ollama, LM Studio, vLLM) crashed the model process or ran out of memory. The server responds with code 500–599. In "More details" - his answer in full:
500 Internal Server Error, 503 Service Unavailable, overloaded_error, api_error or the words Pulse “the model server responded with an error.” The key and the request itself usually have nothing to do with it.
What to do
- Wait a minute and click “Repeat turn” in the error card - in the session feed or in the task. Pulse will send the same request again, with a new turn.
- Local model: check that the server is running and the model is loaded (Ollama or LM Studio window, address in “Settings → Models”). If the server reports low memory, choose a smaller model.
- Error repeats - select another model in the “Model” chip or another agent.
When to write to support
Write to us if the model server is running (another program is receiving responses from it) and Pulse is getting an error over and over again. Click “Send report” in the error card - a report form will open, where the code and server response have already been entered, the secrets are hidden in them. Add which model and which server you chose, and click “View and Send”: the window will show exactly what is left - check and click “Submit”. If the cloud is turned off, the button is called “View and save” - send the resulting file to us in any way.