Step 6

Choose Your AI Models

Select models based on your RAM. Always install nomic-embed-text first — it's required for document search.

bash — always install this first

# Required for all setups — enables document search (RAG)
ollama pull nomic-embed-text

Pick Your RAM Tier

8 GB RAM

Basic

Entry level — light daily use

llama3.2

Primary model2 GB

mistral

EU alternative4 GB

✅Basic AI chat

✅Document search

⚡Limited reasoning

⚡10–30 sec responses

💡Use cloud AI for heavy tasks

16 GB RAM

Standard

Good for daily use

llama3.1:8b

Chief of Staff5 GB

llama3.2

Fast tasks2 GB

✅Good reasoning

✅Solid document handling

✅5–15 sec responses

32 GB RAM ⭐ Recommended

Sweet Spot

Best balance of quality & speed

qwen2.5:14b

Chief of Staff9 GB

llama3.1:8b

Mid-tier5 GB

llama3.2

Fast tasks2 GB

✅Excellent orchestration

✅Strong function calling

✅3–10 sec responses

✅~15 GB free for OS + app

64 GB RAM

Power

Near GPT-4 quality, fully local

qwen2.5:14b

Chief of Staff9 GB

llama3.3:70b

Heavy tasks40 GB

llama3.2

Fast tasks2 GB

✅GPT-4 class reasoning locally

✅100% private — nothing leaves

⚡70B slow on CPU — needs GPU

Install Commands by Tier

bash — 8 GB RAM setup

ollama pull nomic-embed-text   # always required
ollama pull llama3.2           # 2 GB — fast small model
ollama pull mistral            # 4 GB — EU alternative (optional)

# .env settings:
OLLAMA_MODEL=llama3.2
OLLAMA_FAST_MODEL=llama3.2
OLLAMA_EMBED_MODEL=nomic-embed-text

bash — 16 GB RAM setup

ollama pull nomic-embed-text   # always required
ollama pull llama3.1:8b        # 5 GB — Chief of Staff
ollama pull llama3.2           # 2 GB — fast tasks

# .env settings:
OLLAMA_MODEL=llama3.1:8b
OLLAMA_FAST_MODEL=llama3.2
OLLAMA_EMBED_MODEL=nomic-embed-text

bash — 32 GB RAM setup (recommended)

ollama pull nomic-embed-text   # always required
ollama pull qwen2.5:14b        # 9 GB — Chief of Staff
ollama pull llama3.1:8b        # 5 GB — mid-tier
ollama pull llama3.2           # 2 GB — fast tasks

# .env settings:
OLLAMA_MODEL=qwen2.5:14b
OLLAMA_FAST_MODEL=llama3.2
OLLAMA_EMBED_MODEL=nomic-embed-text

bash — 64 GB RAM setup

ollama pull nomic-embed-text   # always required
ollama pull qwen2.5:14b        # 9 GB — Chief of Staff
ollama pull llama3.3:70b       # 40 GB — heavy reasoning (needs GPU)
ollama pull llama3.2           # 2 GB — fast tasks

# .env settings:
OLLAMA_MODEL=qwen2.5:14b
OLLAMA_FAST_MODEL=llama3.2
OLLAMA_EMBED_MODEL=nomic-embed-text

← GPU Drivers PostgreSQL →