OpenCode Desktop 1.18.2 sends messages but models never reply after update

[alexandre-leng]

Description

Since updating to OpenCode Desktop 1.18.2 on Windows, I can no longer receive any response from the model.

I can type and send a message, but OpenCode remains stuck and no answer is displayed. The API key is correctly configured and was working before the update. I also checked that the configured API key is valid.

There is no visible error in the interface. The request appears to be sent, but the API or model never returns a response.

This issue started immediately after updating OpenCode Desktop.

Plugins

No

OpenCode version

1.18.2 desktop

Steps to reproduce

Steps to reproduce

Open OpenCode Desktop 1.18.2 on Windows.

Configure a valid API key in the provider settings.

Open a new conversation.

Select a model.

Enter any message, for example: “Hello”.

Send the message.

The message is submitted, but no response is ever received from the model.

Expected behavior

The selected model should return a response normally.

Actual behavior

The message is sent, but the conversation stays without any model response. There is no useful error message shown in the application.

Screenshot and/or share link

opencode-debug-20260716T084504.zip

Operating System

windos 11

Terminal

desktop app on windows

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• Desktop stuck in "thinking" mode - sidecar stream never completes (v1.17.18, Windows) #36370: Desktop stuck in "thinking" mode on Windows — sidecar stream starts but never completes, all models affected (very similar symptoms)
• Desktop: Unable to chat for hours after auto-update #29862: Desktop unable to chat for hours after auto-update
• Last couple of releases have serious hanging #37241: Serious hanging in recent releases on Windows 11

If your issue is the same as one of these, please add a comment there with any additional details (e.g. your debug zip) rather than opening a new report. If it is distinct, feel free to keep this open.

[amcique]

yeah they fked up pretty bad:

Error from provider (Console): Model deepseek-v4-flash-free is not supported on the full model list. Use GET /inference/openai/v1/models to list available models.