Chat messages not sending

If your message appears stuck or no response arrives after pressing Send, the most common cause is a stopped gateway. Go to Settings > Gateway and start the gateway, then resend your message.

If the gateway is already running, work through the causes below in order.

Gateway is offline

The local gateway must be running to relay messages to the agent. Without it, nothing you type can reach the agent.

1. Open Settings > Gateway.
2. If the status shows Stopped, click Start.
3. Wait for the status to change to Running, then resend your message.

Agent is not connected

If the agent status shows Disconnected, messages cannot be delivered even when the gateway is running.

Follow the steps in Agent shows Disconnected to restore the connection, then retry.

No AI model configured

An agent with no model assigned cannot process messages.

1. Open the agent in My Agents.
2. Go to the Profile tab and verify a model is selected.
3. If no model is set, click Edit and choose a model from the dropdown.

API key invalid or out of credits

The LLM provider rejected the request. The chat input area usually shows the provider's own error code.

1. Look for an error like "Invalid API key" or "Insufficient quota" in the chat.
2. Open the agent's profile, click Edit, and verify the API key for the selected provider is correct and active.
3. Check your account balance or usage limits on the provider's dashboard.

Rate limit hit

If you have sent many messages in a short period, the provider may have throttled your requests temporarily.

Wait a few minutes, then resend your message. You can also open the agent's profile and switch to a different configured model to continue immediately.

Network issue reaching the provider

Your machine must be able to reach the LLM provider's API over the internet. A VPN, firewall, or dropped connection can silently block requests.

1. Confirm you have a working internet connection.
2. Check whether a VPN or firewall policy might be blocking the provider's domain (for example, api.openai.com).
3. Disable the VPN temporarily to test, then re-enable it and add an exception if needed.