AI Hardware Compatibility Matrix

Filter laptops and workstations by architecture (ARM vs x86), minimum memory, and local runner (Ollama, LM Studio, AnythingLLM) to verify compatibility with 7B and 8B parameter open-weights models.

About these numbers: token speeds are estimates based on published specs (mainly memory bandwidth) and public community results, not our own lab measurements. Real performance varies with runtime version, quantization, context length and settings.