← sources

GLM 5.3 Flash just stops during long reasoning (ollama cloud old max plan)

source: reddit (reddit:ollama)published 2026-09-25 12:12ingested 2026-09-26 02:38

retained text

This keeps happening and has been happening for over a month now. Since ollama has zero support I'm hoping someone else has figured out a work around for the issue. I'm using the latest version of opencode (upgraded this morning) and the public https endpoint. 

The following is a screenshot showing an example where the session just stalls while they're in the middle of thinking:

https://preview.redd.it/b7zlqcc1qnrh1.png?width=994&format=png&auto=webp&s=a916ded8e07662b489f04c319d6bfb6d95a57cb4

This is extremely derailing because it happens at a really absurd rate. This session alone (which has a well scoped task with clear acceptance criteria) has stalled 4 times now. They've never actually gotten started and have just read a few files and then thought the rest of the time. 

https://preview.redd.it/nxw5i0kmqnrh1.png?width=1021&format=png&auto=webp&s=d7384c8522aa9e131daa240d13799069fdd9d2f4

This session has lasted for about 2 hours now when it should have taken \~10-15 minutes. This delay is entirely due to this issue and this issue doesn't seem to exist with other providers. I tested several providers on openrouter and tested synthetic\[.\]new and the issue only seems to exist with ollama cloud. 

Has anyone else figured out a work around for this issue? I'm actively looking for another provider so I can move away from ollama cloud but I'm stuck using it for now until I do. 

document fetched 2026-09-26 00:41 · canonical https://reddit.com/r/ollama/comments/1wpupjl/glm_53_flash_just_stops_during_long_reasoning