Model Quant Measured Estimated RAM used App Source Date
Mistral 7B Mistral 7B Q4 128 tok/s 61 tok/s +110% 4.3 GB LM Studio editorial 2026-02-10
Qwen 2.5 7B Qwen 2.5 7B Q4 120 tok/s 62 tok/s +94% 4.6 GB Ollama community 2026-02-12
Phi-4 Mini Phi-4 Mini Q8 118 tok/s 65 tok/s +82% 4.6 GB LM Studio editorial 2026-02-10
Gemma 3 12B Gemma 3 12B Q4 74 tok/s 53 tok/s +40% 7.2 GB LM Studio editorial 2026-03-08