Installation and pulling models

On MacOS v26 (on MacBook Pro M5)

Ollama itself

curl -fsSL https://ollama.com/install.sh | sh

Models

In use

Best quality: the Apple Silicon (MLX) build (~18 GB)

ollama pull qwen3.8:27b-mlx

Fast: MoE daily driver (~23 GB)

ollama pull qwen3.6:35b-a3b

the official qwen3.6:35b-a3b tag is a 23GB Q4_K_M. That's tight on 32 GB, so if memory gets squeezed, pull a smaller community IQ4_XS instead (~17.7GB):

ollama pull hf.co/bartowski/Qwen_Qwen3.6-35B-A3B-GGUF:IQ4_XS

Small model: for autocomplete

ollama pull qwen3.5:9b

Tested

good, but too big, calculations are long (40'--4h), almost can not predict workflow pace (timing). Sometimes takes too much time for smaller tasks.

ollama pull qwen3-coder:30b

2026-10-05:

Today I have decided to go with (relatively) smaller models, than earlier

ollama rm qwen3.6:35b-a3b
ollama rm qwen3.8:27b-mlx
ollama pull ministral-3:14b
ollama pull ornith:9b
ollama list
ollama pull qwen3:14b
ollama rm qwen3.5:9b



anton-w@ant1mbp7 ~ % ollama list
NAME                    ID              SIZE      MODIFIED
qwen3:14b               bdbd181c33f2    9.3 GB    2 hours ago
ornith:9b               a75697c14589    5.6 GB    2 hours ago
ministral-3:14b         4760c35aeb9d    9.1 GB    2 hours ago
phi4:14b                ac896e5b8b34    9.1 GB    2 hours ago
gemma4:26b              001e5dafc3c7    18 GB     3 days ago
gpt-oss:20b             17052f91a42e    13 GB     6 days ago
devstral-small-2:24b    24277f07f62d    15 GB     6 days ago