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
I'm using the big pickle model, and giving it a prompt, but it ALWAYS randomly stops the response. No error gets thrown, it just plays the session complete sound, and stops responding. sometimes it stops while thinking, and sometimes it stops before it even starts to process the prompt and do anything. I tried using the terminal version too, and the same issue happens there. This has genuinely been annoying me for the past 2-3 days, and I use OpenCode daily. If anyone could look into this, it would be much appreciated.
Plugins
The only plugin I'm using is openqueue
OpenCode version
desktop 1.17.11
Steps to reproduce
Literally input any prompt
Screenshot and/or share link
No response
Operating System
Windows 10
Terminal
No response
Reacted by Oleksandr Vanilchuk, Fei Deng, Marlon U. Marcello, Jonathan, Andreas Wäscher and Josh Wong
I dont know how many times I wrote "continue" today. maybe more than 20 times. on MiMo. opencode version 1.18.19. it stops every 4 minutes or so. on an average it stops every 5 minutes.
Keeps doing it to me all the time. It just stops mid-sentence/reasoning for NO reason. I can't run a plan or any long-horizon coding without being at the pc because it randomly stops during reasoning.
Sometimes it just doesn't do it, but most of the time does.
Also it's stubborn against using subagents even if I ask to. My Qwen 3.8 instance has low context about 65k, and I specifically ask to break tasks in small chunks to let alone agents with full context break those small tasks and report to main session.
Same for me, I first found this to happen in OpenChamber (reported here) with Ox-Alpha and some local Qwen 3.8 models, but found out that the underlying OpenCode seems to have the issue:
Activity
github-actions commentedon Jun 29, 2026
This issue might be a duplicate of existing issues. Please check:
Please review those issues and add any additional context (such as logs or screenshots) that might help distinguish your case if it's different.
nexxeln commentedon Jun 30, 2026
can you export the session and attach it please?
dattarohu-coder commentedon Jul 1, 2026
https://opncd.ai/share/cOed0ly7
I recently updated to the latest update, it now gets a little farther, but in the shared link, you can see that it clearly gets cut off while thinking
Elixir-Piloting commentedon Aug 18, 2026
yeah, and as a go subsciber every time i send a "continue" that is getting counted as a message and deducted from my quota
omani commentedon Aug 21, 2026
I dont know how many times I wrote "continue" today. maybe more than 20 times. on MiMo. opencode version 1.18.19. it stops every 4 minutes or so. on an average it stops every 5 minutes.
omani commentedon Aug 22, 2026
absolute shitshow right now.
Greencus commentedon Aug 22, 2026
this is also constantly happening for me
dynched commentedon Aug 31, 2026
seeing the same behavior
paciox commentedon Sep 7, 2026
Keeps doing it to me all the time. It just stops mid-sentence/reasoning for NO reason. I can't run a plan or any long-horizon coding without being at the pc because it randomly stops during reasoning.
Sometimes it just doesn't do it, but most of the time does.
Also it's stubborn against using subagents even if I ask to. My Qwen 3.8 instance has low context about 65k, and I specifically ask to break tasks in small chunks to let alone agents with full context break those small tasks and report to main session.
It ignores it and doesn't spawns any subagent.
awaescher commentedon Sep 16, 2026
Same for me, I first found this to happen in OpenChamber (reported here) with Ox-Alpha and some local Qwen 3.8 models, but found out that the underlying OpenCode seems to have the issue:
awaescher commentedon Sep 16, 2026
I found out that in every chat that got cut off, the output tokens were exactly at 32000 tokens:
Increasing the max output tokens for oMLX in my case did not work, so it seems to be an OpenCode limit.
awaescher commentedon Sep 16, 2026
Okay, this led me to a whole bunch of related issues but it all boils down to this: #1735
Setting
OPENCODE_EXPERIMENTAL_OUTPUT_TOKEN_MAXactually works. Definingmax_tokensin the OpenCode provider config does not work (yet).thdxr commentedon Oct 11, 2026
Thanks for the report! OpenCode 2.0 is now out and development is focused on V2, so we're closing issues filed against OpenCode 1. Please upgrade (
opencode upgrade --major, details: https://anoma.ly/notes/opencode-2-0), and if this is still relevant on 2.0, reopen this issue or open a new one with your 2.0 version.— from 𝕺𝖕𝖊𝖓𝕮𝖔𝖉𝖊