Model Quant Measured Estimated RAM used App Source Date
Gemma 3 1B Gemma 3 1B Q8 52 tok/s 31 tok/s +68% 1.5 GB Ollama editorial 2026-03-15
Phi-4 Mini Phi-4 Mini Q4 28 tok/s 9 tok/s +211% 2.4 GB LM Studio editorial 2026-03-15