You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Opencode Web: Thinking is not differentiable from main #14340
I am not able to differentiate between thinking and main response from agents in the web UI after the 1.2.6 update
There is no collapse for the thinking and all the thinking and main text is shown on same screen,
The only differences are -
thinking has slightly light color than main text
font size is little smaller for the thinking text, one pixel difference maybe
Screenshot -
Plugins
No response
OpenCode version
1.2.6
Steps to reproduce
start opencode with opencode web
Ask any quesiton in the chat which might trigger thinking
The main issue is not only that thinking/reasoning takes space. The harder problem is that the actual final answer / conclusion is still too easy to miss because the distinction relies mostly on subtle styling.
I use OpenCode regularly, and this still does not become intuitive with familiarity. I often have to stop and visually scan to figure out where the real answer starts versus where the reasoning/explanation ends.
A stronger visual boundary would help a lot, for example:
a dedicated Final Answer / Conclusion heading
a clear divider between reasoning and final content
a more obvious container/background difference
keeping reasoning collapsible, but also making the final answer visually anchored
In other words, collapsible reasoning would help, but it does not fully solve the problem unless the final answer is also made much easier to identify at a glance.
Activity
GravityPoet commentedon Apr 6, 2026
Still hitting this from a UX perspective.
The main issue is not only that thinking/reasoning takes space. The harder problem is that the actual final answer / conclusion is still too easy to miss because the distinction relies mostly on subtle styling.
I use OpenCode regularly, and this still does not become intuitive with familiarity. I often have to stop and visually scan to figure out where the real answer starts versus where the reasoning/explanation ends.
A stronger visual boundary would help a lot, for example:
In other words, collapsible reasoning would help, but it does not fully solve the problem unless the final answer is also made much easier to identify at a glance.
github-actions commentedon Jul 4, 2026
To stay organized issues are automatically closed after 60 days of no activity. If the issue is still relevant please open a new one.
feat(session-ui,app): collapsible reasoning cards for desktop/web