[ALL TASKS FAILED] Incapable of do correctly a single task

[cgonIT]

I'm starting to get frustrated. Here's a screenshot of my current setup:

ℹ️

• Windows 11 with 64GB RAM
• nVidia RTX3070ti
• Docker Desktop. Inside docker, 2 composes: Ollama (web and ollama itself), and Bytebot in litlellm-proxy mode deployed.

✅OLLAMA:

• It doesn't matter what LLM installed, they all work perfect. First interaction takes a little bit longer, next ones almos suddenly.
• Models right now:

🚨BYTEBOT:

• As mentioned, docker version with the litlellm-proxy compose:

  • model_name: gpt-oss
    litellm_params:
    drop_params: true
    model: ollama/gpt-oss:latest
    api_base: http://host.docker.internal:11434
  • model_name: deepseek-r1
    litellm_params:
    model: ollama/deepseek-r1:latest
    api_base: http://host.docker.internal:11434`

⚠️MY PROBLEM:
It's impossible to get a simple task like opening Firefox and opening Google to work. Or even opening a text editor, typing a word, and saving it. EVERY SINGLE tasks end within a few seconds, displaying “Task Failed.”

There is communication between Bytebot and Ollama because I'm constantly checking both container logs (Bytebot Agent and Ollama, below screenshot of both logs):

Bytebot-agent node execution logs:

Ollama node execution logs:

I tried doing search at google, here in the issues at git, in the Bytebot main site... but totally frustrated.
Can someone help me to solve this problems? Thanks a million in advanced

[zhiming1996]

me, too

[dmitrymailk]

try to use lm studio https://lmstudio.ai/ . it works sometimes with qwen3-vl.

[edusanzio]

me too
mine can open tha requested webpage, but keeps stuck and cannot click the e-mail textbox, so frustrating. The idea is good but it's just not working right now