give me a clue? | Nous Research | Page 1

undone stratus May 3, 2026, 7:59 PM

#

I think it tried to load the mode.
But the model is not loaded.
Instead of loading the model, it just failed.

But in OpenClaw, when the model not loaded, it will make LM Studio load it.

After the prompt processing and the token generation are done, and 10minutes pass, the TTL unloaded the model and free the RAM.

violet parcel May 3, 2026, 8:00 PM

#

what is the API error?

#

400?

#

404?

#

200?

#

500?

#

402?

#

and does the last couple of the error suggest what to do?

undone stratus May 3, 2026, 8:17 PM

#

I've added the lmstudio via cli right.

But this is the LM Studio log respond, it did show that hermes try to access it, but it "refuse"

2026-05-04 03:56:48 [ERROR]
Unexpected endpoint or method. (POST /v1/chat/completions). Returning 200 anyway
2026-05-04 04:16:23 [DEBUG]
Received request: GET to /v1/models
2026-05-04 04:16:23 [ERROR]
Unexpected endpoint or method. (GET /v1/models). Returning 200 anyway

violet parcel May 3, 2026, 8:20 PM

#

undone stratus I've added the lmstudio via cli right. But this is the LM Studio log respond, i...

did you enable CORS in LMStudio?

undone stratus May 3, 2026, 8:21 PM

#

Of course, this is the same LMStudio that accessed by my OpenClaw agent heartbeats.

violet parcel May 3, 2026, 8:49 PM

#

Ok so a 200 error means malformed API so maybe it stops ar \v1 or \completions

#

Check it

#

TL;DR: LM Studio is rejecting Hermes' API calls because its Local Server mode isn't properly enabled — you're in Chat UI mode, not Server mode. Open LM Studio, click the "<->" (Local Server) tab, load a model, click "Start Server", then configure Hermes with hermes config set model.base_url http://localhost:1234/v1. On Termux/Android? LM Studio won't run locally — use a desktop's IP or switch to Ollama.

Full details attached.

#give me a clue?