ollama-modelfiles

Ollama Modelfile collection for local AI coding workflows, optimized for two machines.

Hardware

Machine	Chip	RAM	VRAM
MacBook	Apple M3	24 GB Unified Memory	—
PC	NVIDIA RTX 5090	64 GB RAM	32 GB VRAM

Models

Model	Type	Size (Q4_K_M)	Mac ctx	PC ctx
gemma4:26b-a4b-it-q4_K_M	MoE	~18 GB	16384	65536
gemma4:31b-it-q4_K_M	Dense	~21 GB	8192	32768
qwen3.5:27b-q4_K_M	Dense	~17 GB	8192	32768
qwen3.6:35b-a3b-q4_K_M	MoE	~24 GB	—	65536

Setup

Mac Setup

# Pull models
ollama pull gemma4:26b-a4b-it-q4_K_M
ollama pull gemma4:31b-it-q4_K_M
ollama pull qwen3.5:27b-q4_K_M

# Create custom configs
ollama create gemma4-26b-coding -f ./mac-m3-24gb/gemma4-26b-a4b-it-q4_K_M.txt
ollama create gemma4-31b-planning -f ./mac-m3-24gb/gemma4-31b-it-q4_K_M.txt
ollama create qwen3.5-27b-planning -f ./mac-m3-24gb/qwen3.5-27b-q4_K_M.txt
ollama list

PC Setup

# Pull models
ollama pull gemma4:26b-a4b-it-q4_K_M
ollama pull gemma4:31b-it-q4_K_M
ollama pull qwen3.5:27b-q4_K_M
ollama pull qwen3.6:35b-a3b-q4_K_M

# Create custom configs
ollama create gemma4-26b-coding -f ./pc-rtx5090-32gb/gemma4-26b-a4b-it-q4_K_M.txt
ollama create gemma4-31b-planning -f ./pc-rtx5090-32gb/gemma4-31b-it-q4_K_M.txt
ollama create qwen3.5-27b-planning -f ./pc-rtx5090-32gb/qwen3.5-27b-q4_K_M.txt
ollama create qwen3.6-35b-coding -f ./pc-rtx5090-32gb/qwen3.6-35b-a3b-q4_K_M.txt
ollama list

Repository Structure

ollama-modelfiles/
├── README.md
├── mac-m3-24gb/
│   ├── gemma4-26b-a4b-it-q4_K_M.txt
│   ├── gemma4-31b-it-q4_K_M.txt
│   └── qwen3.5-27b-q4_K_M.txt
├── pc-rtx5090-32gb/
│   ├── gemma4-26b-a4b-it-q4_K_M.txt
│   ├── gemma4-31b-it-q4_K_M.txt
│   ├── qwen3.5-27b-q4_K_M.txt
│   └── qwen3.6-35b-a3b-q4_K_M.txt
└── scripts/
    └── ollama-network-expose.ps1

Parameters

Parameter	Value	Description
num_ctx	varies by model/hardware	Maximum context length in tokens
num_predict	2048 / 4096	Maximum response length
temperature	0.2 (coding) / 0.5 (planning)	Coding = deterministic output; Planning = more creative reasoning
repeat_penalty	1.1	Prevents repetitive outputs

VS Code Copilot Chat

After setup, select the desired model in Copilot Chat:

• gemma4-26b-coding
• gemma4-31b-planning
• qwen3.6-35b-coding (PC only)
• qwen3.5-27b-planning

Scripts

Script	Platform	Description
scripts/ollama-network-expose.ps1	Windows (PowerShell)	Exposes the Ollama API on the local network

# Run from repo root in an elevated PowerShell session
powershell -ExecutionPolicy Bypass -File .\scripts\ollama-network-expose.ps1