whatcani.run on an M3 Pro with 36 GB RAMM3 Pro with 36 GB RAM?

Find the best models and how to run them locally, based on real data from 31,372,160 tokens across 6,141 trials from 239 people.Find the best models and how to run them locally, based on real data from tokens across trials from people.

Models

~ $ ▌

Models

Results include trials with 4,096 input tokens and 1,024 output tokens only.

								Actions
Qwen3.5-9B Source	llama.cpp	15.8 tok/s	197.6 tok/s	20.34 sec	2.93 GB 8%	101 run	Runs ok

			Actions
Qwen3.5-9B Source	15.8 tok/s	Runs ok

								Actions
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source

			Actions
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source
Source