Model Quant Measured Estimated RAM used App Source Date
Qwen 2.5 0.5B Qwen 2.5 0.5B Q8 75 tok/s 45 tok/s +67% 0.6 GB LM Studio editorial 2026-03-20
Gemma 3 1B Gemma 3 1B Q4 38 tok/s 24 tok/s +58% 0.8 GB Ollama editorial 2026-03-20