[BUG] Not going past Google home screen

[Dogway]

Title

Used several APIs but Agent-E doesn't perform any actions.

Description

I first tested with openrouter API, using tool models like Llama 3.2 3B Instruct (free) or Gemini 1.5 free without success.
Since for some reason openrouter was asking for credits on the terminal, I enabled a local server with lmstudio and served Llama 3.2 3B Instruct from there, without success again. I know the local server works because I tested on OpenWebUI but in Agent-E chatbot it only replies with the instructions and stops there. I attach the log.

Steps to Reproduce

1. Setup the API Base_URL and KEY and model name in agents_llm_config.json
2. Launch Agent-E
3. Ask for an action

Expected Behavior

I expect the agent, which is fed with a tool model to perform the action prompted

Actual Behavior

It replies and stays idle

Screenshots

Environment

Windows 11, Chrome latest

Additional Context

Attached a log
Agent-E - error log.txt

[teaxio]

I think this is an issue with the model u are using llama-3.2-3b-instruct I do not think these small models can handle this.

[Dogway]

Does it need to follow a certain schema? What models aside from OpenAI's do you recommend that works?