OpenHands

Has anyone gotten the CLI or GUI with CLI (openhands serve) working well with a model running from Ollama?

For context I have it up and running locally but wondering if anyone is using it consistently in practice

I don’t, currently, but if it works well for you I’m sure we’d love to know! I believe that sooner or later, local models will get to the point of supporting incredible stuff :pray:

Its working _okay_ but not great. I don't have any standout hardware but qwen3-coder:30b is able to run slowly

Did you follow the instructions here? <https://docs.openhands.dev/openhands/usage/llms/local-llms#quickstart-running-openhands-with-a-local-llm-using-lm-studio>

I think with a model that’s as strong as qwen3-coder:30b, probably “ok but not great” is the best you’ll get. qwen3-coder-next might start getting more decent results

But if the results seem OK, then I guess it’s probably working, and the rest is just a model skill issue

When things aren’t working, the model will do stuff like not make any tool calls at all

Yep! That doc with a bit of fiddling got things working. I'll try some other models and see how things play out