No Response Requested #1901

Open
opened 2026-02-16 17:33:08 -05:00 by yindo · 25 comments
Owner

Originally created by @JoshuaRileyDev on GitHub (Oct 1, 2025).

Originally assigned to: @fwang on GitHub.

So I have an issue with using GLM 4.6. I've added it correctly but I get:

"No response requested."

Also my previous attempt was stuck on generating for a while and couldn't see any code being generated so was worried it was just stuck so aborted.

Image
Originally created by @JoshuaRileyDev on GitHub (Oct 1, 2025). Originally assigned to: @fwang on GitHub. So I have an issue with using GLM 4.6. I've added it correctly but I get: "No response requested." Also my previous attempt was stuck on generating for a while and couldn't see any code being generated so was worried it was just stuck so aborted. <img width="567" height="165" alt="Image" src="https://github.com/user-attachments/assets/45825119-dfbe-45e0-a271-573b7eca6142" />
yindo added the model-problem label 2026-02-16 17:33:08 -05:00
Author
Owner

@github-actions[bot] commented on GitHub (Oct 1, 2025):

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

  • #2883: Same GLM-4.6 model support issue (closed, but directly mentions GLM-4.6 not being available)
  • #1880: GLM 4.5 string issue with similar error symptoms and model problems
  • #2512: Opencode stuck in "generating" - same stuck behavior you experienced
  • #2557: GLM-4.5 API parameter issues that cause hanging/errors (closed, but shows similar GLM model problems)
  • #2137: Message generation interrupted, stuck at "Generating" - very similar symptoms

Feel free to ignore if none of these address your specific case.

@github-actions[bot] commented on GitHub (Oct 1, 2025): This issue might be a duplicate of existing issues. Please check: - #2883: Same GLM-4.6 model support issue (closed, but directly mentions GLM-4.6 not being available) - #1880: GLM 4.5 string issue with similar error symptoms and model problems - #2512: Opencode stuck in "generating" - same stuck behavior you experienced - #2557: GLM-4.5 API parameter issues that cause hanging/errors (closed, but shows similar GLM model problems) - #2137: Message generation interrupted, stuck at "Generating" - very similar symptoms Feel free to ignore if none of these address your specific case.
Author
Owner

@rekram1-node commented on GitHub (Oct 1, 2025):

Hmm I need to signup for the plan to test this better, can you share the session by chance?

It almost seems like the model just "said" that?

@rekram1-node commented on GitHub (Oct 1, 2025): Hmm I need to signup for the plan to test this better, can you share the session by chance? It almost seems like the model just "said" that?
Author
Owner

@JoshuaRileyDev commented on GitHub (Oct 1, 2025):

I've switched to my CC plan atm but can switch back and see if I can replicate again then share the session

@JoshuaRileyDev commented on GitHub (Oct 1, 2025): I've switched to my CC plan atm but can switch back and see if I can replicate again then share the session
Author
Owner

@slipstr28 commented on GitHub (Oct 1, 2025):

I encountered this message ( or only : OK. too after using continue on an alternative routing provider with GLM 4.6 whom we already had 'stuck in sessions' with GLM 4.5.
looking at the logs, this 'stuck' behaviour or 'weird' message just happens after automated tool pruning.
It happened also under GLM 4.5 for this specific provider, also after a pruning.

These tool pruning did happen without auto compaction --> Session summary, which let me think that the model exhausted its context ( tool use in majority ) without possibility of recovery.

In regard to the anomalies with this specific proxying provider, my hypothesis is that somehow the served model announced 128K available context but had a smaller internal context windows, which exhausted far before the ide might trigger compaction --> Summary.

I was able to reproduce this kind of anomaly by using the Z.ai anthropic endpoint with GLM 4.6 ( setup as a custom provider) and let the context expire by itself with a moderate code or refactor task with edits and slight use of context7.

The log show exactly the tool auto prune before the anomaly appears.
INFO 2025-10-01T18:08:53 +22ms service=session.compaction pruned=10282 total=46436 found

Hope it helps.

@slipstr28 commented on GitHub (Oct 1, 2025): I encountered this message ( or only : `OK. ` too after using continue on an alternative routing provider with GLM 4.6 whom we already had 'stuck in sessions' with GLM 4.5. looking at the logs, this 'stuck' behaviour or 'weird' message just happens after automated tool pruning. It happened also under GLM 4.5 for this specific provider, also after a pruning. These tool pruning did happen without auto compaction --> Session summary, which let me think that the model exhausted its context ( tool use in majority ) without possibility of recovery. In regard to the anomalies with this specific proxying provider, my hypothesis is that somehow the served model announced 128K available context but had a smaller internal context windows, which exhausted far before the ide might trigger compaction --> Summary. I was able to reproduce this kind of anomaly by using the Z.ai anthropic endpoint with GLM 4.6 ( setup as a custom provider) and let the context expire by itself with a moderate code or refactor task with edits and slight use of context7. The log show exactly the tool auto prune before the anomaly appears. INFO 2025-10-01T18:08:53 +22ms service=session.compaction pruned=10282 total=46436 found Hope it helps.
Author
Owner

@rekram1-node commented on GitHub (Oct 1, 2025):

that does help ty

@rekram1-node commented on GitHub (Oct 1, 2025): that does help ty
Author
Owner

@n0cloud commented on GitHub (Oct 2, 2025):

It happened to me after interrupting the agent with esc. Any request after that just returned the "No Response Requested" error.

@n0cloud commented on GitHub (Oct 2, 2025): It happened to me after interrupting the agent with esc. Any request after that just returned the "No Response Requested" error.
Author
Owner

@towry commented on GitHub (Oct 4, 2025):

# Conversation History

---

**User** (*2025-10-05 05:13:37*)

@sage hi

Use the above message and context to generate a prompt and call the task tool with subagent: sage

---

**Assistant** (*2025-10-05 05:13:38*)



No response requested.

---

**User** (*2025-10-05 05:14:11*)

hi

---

**Assistant** (*2025-10-05 05:14:11*)



No response requested.


I am using the zhipu coding plan.

@towry commented on GitHub (Oct 4, 2025): ``` # Conversation History --- **User** (*2025-10-05 05:13:37*) @sage hi Use the above message and context to generate a prompt and call the task tool with subagent: sage --- **Assistant** (*2025-10-05 05:13:38*) No response requested. --- **User** (*2025-10-05 05:14:11*) hi --- **Assistant** (*2025-10-05 05:14:11*) No response requested. ``` I am using the zhipu coding plan.
Author
Owner

@williamhrs commented on GitHub (Oct 6, 2025):

it happens a lot here.

@williamhrs commented on GitHub (Oct 6, 2025): it happens a lot here.
Author
Owner

@ShamanicArts commented on GitHub (Oct 8, 2025):

happens all the time. sometimes it fixes itself by just reminding the model you are requesting a response, & sometimes that session is done lol

@ShamanicArts commented on GitHub (Oct 8, 2025): happens all the time. sometimes it fixes itself by just reminding the model you are requesting a response, & sometimes that session is done lol
Author
Owner

@kierr commented on GitHub (Oct 10, 2025):

@thdxr you were asking for GLM 4.6 feedback, has this quirk been addressed? I've been using Opencode and GLM 4.6 for 6 days and this is the main blocker, along with timeouts, random 40x/50x API errors, and no retries. This is via ZAI Coding Plan. Works better with Claude Code because it more aggressively retries, whereas Opencode craps out fast and idles until you say 'Continue'.

@kierr commented on GitHub (Oct 10, 2025): @thdxr you were asking for GLM 4.6 feedback, has this quirk been addressed? I've been using Opencode and GLM 4.6 for 6 days and this is the main blocker, along with timeouts, random 40x/50x API errors, and no retries. This is via ZAI Coding Plan. Works better with Claude Code because it more aggressively retries, whereas Opencode craps out fast and idles until you say 'Continue'.
Author
Owner

@williamhrs commented on GitHub (Oct 17, 2025):

still happening, sometimes after we have to stop it to change course with ESC, sometimes out of nowhere after a task

@williamhrs commented on GitHub (Oct 17, 2025): still happening, sometimes after we have to stop it to change course with ESC, sometimes out of nowhere after a task
Author
Owner

@bennyzen commented on GitHub (Oct 21, 2025):

it's related to interrupting using the ESC key. confirmed.

and yes, unfortunately it's happening all the time.

@bennyzen commented on GitHub (Oct 21, 2025): it's related to interrupting using the ESC key. confirmed. and yes, unfortunately it's happening all the time.
Author
Owner

@kierr commented on GitHub (Oct 21, 2025):

If anyone is still facing this issue, please report your opencode --version too.

I haven't seen this happen for a while, despite using GLM 4.6 via Coding Plan and pressing ESC often.

Make sure you're using the latest version (bunx opoencode-ai@latest).

@kierr commented on GitHub (Oct 21, 2025): If anyone is still facing this issue, please report your `opencode --version` too. I haven't seen this happen for a while, despite using GLM 4.6 via Coding Plan and pressing ESC often. Make sure you're using the latest version (`bunx opoencode-ai@latest`).
Author
Owner

@bennyzen commented on GitHub (Oct 21, 2025):

v0.15.12

latest. the issue still persists

@bennyzen commented on GitHub (Oct 21, 2025): v0.15.12 latest. the issue still persists
Author
Owner

@david-crespo commented on GitHub (Nov 5, 2025):

Happened in 1.0.25 after I aborted a previous command.

Image
@david-crespo commented on GitHub (Nov 5, 2025): Happened in 1.0.25 after I aborted a previous command. <img width="657" height="527" alt="Image" src="https://github.com/user-attachments/assets/b1e75202-03ea-4f3d-8d6a-f953a924f564" />
Author
Owner

@zRzRzRzRzRzRzR commented on GitHub (Nov 7, 2025):

This question is about whether to use the API (z.ai) version or the open-source model deployed in SGLang or vLLM.

@zRzRzRzRzRzRzR commented on GitHub (Nov 7, 2025): This question is about whether to use the API (z.ai) version or the open-source model deployed in SGLang or vLLM.
Author
Owner

@pnocera commented on GitHub (Nov 14, 2025):

This also happens with cerebras provider with GLM 4.6. So I suppose it's not directly related to z.ai endpoint but to the model itself ?

@pnocera commented on GitHub (Nov 14, 2025): This also happens with cerebras provider with GLM 4.6. So I suppose it's not directly related to z.ai endpoint but to the model itself ?
Author
Owner

@rekram1-node commented on GitHub (Nov 14, 2025):

yeah it seems like it, when I get a chance ill try to look into it more

@rekram1-node commented on GitHub (Nov 14, 2025): yeah it seems like it, when I get a chance ill try to look into it more
Author
Owner

@JosXa commented on GitHub (Nov 20, 2025):

Image

Seems to be specifically in the thinking blocks. On Cerebras, the schema doesn't seem to be followed correctly, and the AI SDK struggles with it.

@JosXa commented on GitHub (Nov 20, 2025): <img width="1070" height="261" alt="Image" src="https://github.com/user-attachments/assets/101f9653-c804-4279-b7e3-8932efd59e97" /> Seems to be specifically in the thinking blocks. On Cerebras, the schema doesn't seem to be followed correctly, and the AI SDK struggles with it.
Author
Owner

@burka commented on GitHub (Nov 21, 2025):

Same here with Cerebras Z.ai GLM-4.6, every now and then it just halts, sometimes very often - a compact does help sometimes, but not always

Image
@burka commented on GitHub (Nov 21, 2025): Same here with Cerebras Z.ai GLM-4.6, every now and then it just halts, sometimes very often - a compact does help sometimes, but not always <img width="487" height="250" alt="Image" src="https://github.com/user-attachments/assets/191ed20a-cc24-43a9-9539-659a344a7c46" />
Author
Owner

@burka commented on GitHub (Nov 21, 2025):

I’ve created a pull request for this. It works on my machine so far :-) https://github.com/sst/opencode/pull/4584

@burka commented on GitHub (Nov 21, 2025): I’ve created a pull request for this. It works on my machine so far :-) https://github.com/sst/opencode/pull/4584
Author
Owner

@JosXa commented on GitHub (Nov 21, 2025):

Need to also cover these more generally:

Image

I'm guessing reasoning ️ but no text ️ or something.

@JosXa commented on GitHub (Nov 21, 2025): Need to also cover these more generally: <img width="702" height="222" alt="Image" src="https://github.com/user-attachments/assets/548b8e7a-b956-4da2-b5a5-732fdf37f27f" /> I'm guessing reasoning ✅️ but no text ❌️ or something.
Author
Owner

@towry commented on GitHub (Nov 21, 2025):

I’ve created a pull request for this. It works on my machine so far :-) https://github.com/sst/opencode/pull/4584

Please do not merge this, it is footgun

@towry commented on GitHub (Nov 21, 2025): > I’ve created a pull request for this. It works on my machine so far :-) https://github.com/sst/opencode/pull/4584 Please do not merge this, it is footgun
Author
Owner

@JosXa commented on GitHub (Nov 21, 2025):

While we struggle to fix this, disable reasoning as a workaround:
https://github.com/sst/opencode/issues/3259#issuecomment-3560472815

I find GLM's reasoning pretty useless anyway

@JosXa commented on GitHub (Nov 21, 2025): While we struggle to fix this, disable reasoning as a workaround: https://github.com/sst/opencode/issues/3259#issuecomment-3560472815 I find GLM's reasoning pretty useless anyway
Author
Owner

@TomLucidor commented on GitHub (Jan 19, 2026):

Having the same issues with using local models in https://github.com/cubist38/mlx-openai-server
@JosXa I am surprised it has something to do with parsing the reasoning bits!
@ShamanicArts how are reminders done again? Maybe it's just a "dumb zone" thing?

@TomLucidor commented on GitHub (Jan 19, 2026): Having the same issues with using local models in https://github.com/cubist38/mlx-openai-server @JosXa I am surprised it has something to do with parsing the reasoning bits! @ShamanicArts how are reminders done again? Maybe it's just a "dumb zone" thing?
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#1901