ai-christianson

• Experiences with DS4 on long-lived agents

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 10 comments

• I got fed up with Open WebUI/LibreChat for local LLMs so I made an open source tool to turn my GPU server into an always-on assistant

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 20 comments

• We open-sourced our browser agent sandbox: run arbitrary code from local LLMs without torching your system

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 6 comments

• Aider + langchain: A match made in heaven?

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 13 comments

• RA.Aid v0.10.0 - Web research, interactive chat, and more

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 7 comments

• In 44 lines of code, we have an actually useful agent that runs entirely locally, powered by Qwen3 30B A3B Instruct

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 13 comments

• How to run mlx-optimized models on Apple (gets best tok/sec)

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 2 comments

• Using qwq-32b effectively in coding agents

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 5 comments

• Qwen 32b coder instruct can now drive a coding agent fairly well

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 72 comments

• RA.Aid v0.14.3 released with Sonnet 3.7 support.

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 1 comments

• We added open models support to RA.Aid and need help testing

  Posted by ai-christianson@reddit | LocalLLaMA | View on Reddit | 12 comments