Hacker News Reader

If "small models" is the bar, then you can run inference for ~$50 on Raspberry Pi like hardware. I do that with 1.8b-4b models.

LFM 450M for vision task, QWEN 9B Q4 for Orchestration, this provides a good result.

I actually meant a context window of about 50k which is what you need to run OpenClaw well.