Getting Started
with Ollama

# Step 1: Ensure the server is running
# Windows: It auto-starts via system tray
# Linux:   sudo systemctl start ollama
# macOS:   Open the Ollama app, or run:
ollama serve &

# Step 2: Pull some models
ollama pull llama3.2:1b        # Meta Llama 3.2  — ~1.3 GB
ollama pull gemma3:4b          # Google Gemma 3  — ~2.5 GB
ollama pull deepseek-r1:8b     # DeepSeek R1     — ~4.9 GB
ollama pull qwen3:7b           # Qwen 3          — ~4.4 GB
ollama pull mistral            # Mistral 7B      — ~4.1 GB

# Step 3: List downloaded models
ollama list
# NAME              ID            SIZE      MODIFIED
# llama3.2:1b       a80c4f17...   1.3 GB    just now

# Step 4: Start an interactive session
ollama run llama3.2:1b
>>> Hello! What can you help me with?

# Step 5: In another terminal — check what's loaded
ollama ps

# Step 6: Show model details
ollama show llama3.2:1b

# Step 7: Unload when done
ollama stop llama3.2:1b

# Set environment variables for your user account
setx OLLAMA_HOST "0.0.0.0"
setx OLLAMA_ORIGINS "*"
setx OLLAMA_MODELS "D:\ollama-models"  # Optional: change model storage

# Or via GUI: Settings → search "environment variables"
# → Edit environment variables for your account

# After setting, quit Ollama from system tray and relaunch

# Edit the Ollama service configuration
sudo systemctl edit ollama

# Add these lines under [Service]:
[Service]
Environment="OLLAMA_HOST=0.0.0.0"
Environment="OLLAMA_ORIGINS=*"
Environment="OLLAMA_NUM_PARALLEL=2"

# Save and reload
sudo systemctl daemon-reload
sudo systemctl restart ollama

launchctl setenv OLLAMA_HOST "0.0.0.0"
launchctl setenv OLLAMA_ORIGINS "*"
# Then restart the Ollama app from menu bar

pip install ollama

from ollama import chat
from ollama import ChatResponse

response: ChatResponse = chat(
    model='llama3.2',
    messages=[
        {
            'role': 'user',
            'content': 'Explain microservices vs monolith',
        },
    ],
)

print(response.message.content)

from ollama import chat

stream = chat(
    model='gemma3:4b',
    messages=[{
        'role': 'user',
        'content': 'Write a Python function for binary search'
    }],
    stream=True,
)

for chunk in stream:
    print(chunk['message']['content'], end='', flush=True)

# Benchmark the same prompt across multiple models
from ollama import chat
import time

models = ['llama3.2', 'gemma3:4b', 'mistral']
prompt = "Explain the Observer design pattern with a Python example"

for model in models:
    print(f"\n{'='*50}\nModel: {model}\n{'='*50}")
    start = time.time()
    response = chat(
        model=model,
        messages=[{'role': 'user', 'content': prompt}],
    )
    elapsed = time.time() - start
    print(response.message.content[:500])
    print(f"\n⏱ Response time: {elapsed:.1f}s")

npm install ollama

import ollama from "ollama";

const response = await ollama.chat({
  model: "llama3.2",
  messages: [{
    role: "user",
    content: "Explain async/await in JavaScript",
  }],
});

console.log(response.message.content);

# 🪟 Windows — Restart from system tray or Start Menu
# Right-click the Ollama tray icon → Quit, then relaunch
# Or run manually:
ollama serve

# 🐧 Linux
sudo systemctl start ollama

# 🍎 macOS — open the Ollama app from Applications, or:
ollama serve

# Check processor allocation
ollama ps
# PROCESSOR column: "100% GPU" = good  |  "100% CPU" = slow

# If on CPU, try a smaller model
ollama run llama3.2:1b

# Check NVIDIA GPU memory
nvidia-smi

# Check system RAM
free -h          # Linux / macOS
systeminfo        # Windows (look for "Available Physical Memory")

# 1. Stop other running models first
ollama stop llama3.2

# 2. Use a smaller variant
ollama run llama3.2:1b    # instead of 3b
ollama run gemma3:1b      # instead of 4b

# 3. Remove unused models to free disk space
ollama rm deepseek-r1:8b

# 4. Audit what's in memory
ollama ps

# Open log files via Run dialog (Win+R):
%LOCALAPPDATA%\Ollama\app.log      # GUI app log
%LOCALAPPDATA%\Ollama\server.log   # Server log

# Or view in PowerShell:
Get-Content "$env:LOCALAPPDATA\Ollama\server.log" -Tail 50

# Stream live logs
journalctl -u ollama.service -f

# View last 50 lines
journalctl -u ollama.service -n 50