cerebro-vip INEMA.CLUB
início › INEMA.LLMS

Guia completo para rodar Gemma 4 localmente com Ollama e conectá-lo…

INEMA.LLMS · 2026-04-07 · ~4 min · ver no Telegram ↗

INEMA

mas com openai:⌗

from openai import OpenAI

client = OpenAI(
base_url="localhost:11434/v1 ↗",
api_key="ollama" # qualquer valor
)

response = client.chat.completions.create(
model="gemma4:e4b",
messages=[{"role": "user", "content": "oi"}]
)

COMO INSTALAR EM UM SERVIDOR DO ZERO

  1. Atualizar o sistema sudo apt update && sudo apt upgrade -y

  2. Instalar Docker curl -fsSL get.docker.com ↗ | sh sudo systemctl enable --now docker sudo usermod -aG docker $USER newgrp docker

  3. Instalar drivers Nvidia (se tiver GPU) sudo apt install -y ubuntu-drivers-common sudo ubuntu-drivers autoinstall sudo reboot

  4. Instalar Nvidia Container Toolkit (para Docker usar a GPU) curl -fsSL nvidia.github.io ↗ | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg curl -s -L nvidia.github.io ↗ | \ sed 's#deb https://#deb ↗ [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g ↗' | \ sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list sudo apt update && sudo apt install -y nvidia-container-toolkit sudo nvidia-ctk runtime configure --runtime=docker sudo systemctl restart docker

  5. Subir Ollama + Open WebUI

Cria a pasta e o compose: mkdir ~/ollama && cd ~/ollama

Cria o docker-compose.yml: services: ollama: image: ollama/ollama:latest container_name: ollama ports: - "11434:11434" volumes: - ollama-data:/root/.ollama environment: - OLLAMA_HOST=0.0.0.0 - OLLAMA_KEEP_ALIVE=24h restart: unless-stopped deploy: resources: reservations: devices: - driver: nvidia count: all capabilities: [gpu]

open-webui:
  image: ghcr.io/open-webui/open-webui:main
  container_name: open-webui
  ports:
    - "3000:8080"
  extra_hosts:
    - "host.docker.internal:host-gateway"
  environment:
    - OLLAMA_BASE_URL=http://host.docker.internal:11434
  volumes:
    - open-webui-data:/app/backend/data
  restart: unless-stopped

volumes: ollama-data: name: ollama-data open-webui-data: name: open-webui-data

Sobe: docker compose up -d

  1. Baixar um modelo docker exec ollama ollama pull gemma4:e4b

Acessa em IP-DO-SERVIDOR:3000 ↗

Aqui está a versão traduzida para português, já pronta em Markdown 👇

  • 🚀 Ollama + Gemma 4 — Guia CompletoExecute o Claude Code com o novo Gemma 4 do Google — 100% grátis, privado e local
  • 📌 Visão GeralConfigure o Gemma 4 localmente com Ollama e conecte ao Claude Code→ Sem chaves de API→ Sem custos→ Sem internet
  • ⚡ Principais VantagensRecursoBenefício💰 Custo$0 para sempre🔒 PrivacidadeTudo roda localmente⚡ VelocidadeSem latência de rede🧠 CapacidadeÓtimo para tarefas do dia a dia🧰 RequisitosmacOS / Linux / WindowsOllama ≥ 0.20.08GB–20GB de RAMClaude Code (npm install -g @anthropic-ai/claude-code)🛠️ Guia de Instalação1. Instalar Ollama```undefined

brew install ollama

ou⌗

curl -fsSL ollama.com/install.sh ↗ | sh

  1. Iniciar o servidorundefined ollama serve

  2. Baixar o modeloModeloRAMUsoe2b~4GBMáquinas fracase4b~8GBPadrão26b~14GBWorkstation31b~19GB🚀 Melhorundefined ollama pull gemma4:31b

  3. Testar o modeloundefined ollama run gemma4:31b

  4. Conectar ao Claude Codeundefined ollama launch claude

⚡ Início Rápido (copiar e colar)undefined brew install ollama && ollama serve & ollama pull gemma4:31b ANTHROPIC_BASE_URL=localhost:11434 ↗ ANTHROPIC_AUTH_TOKEN=ollama ANTHROPIC_MODEL=gemma4:31b claude

🧪 Comandos úteisundefined ollama list ollama ps ollama stop gemma4:31b ollama rm gemma4:31b ollama --version

🧠 Estratégia de usoUse Gemma 4 paraUse Claude (cloud) paraCódigo simplesSistemas complexosCorreção de bugsArquiteturaBoilerplateCódigo crítico🎯 O que você ganha✅ Código com IA grátis✅ 100% offline✅ Uso ilimitado✅ Licença comercial (Apache 2.0)🔗 Recursoshttps://ollama.comhttps://ollama.com/library/gemma4https://docs.anthropic.com/claude/docs/claude-code✨ DicasUse 31b se tiver RAM suficienteUse e4b em notebooksMantenha o Ollama atualizado🏁 ProntoAgora você está rodando IA local para programação sem custo 🚀
Se quiser, posso já te entregar isso como arquivo .md traduzido pra download também 👍

Ollama + Gemma 4: Ultimate Cheat Sheet
> **Run Claude Code with Google's brand-new Gemma 4 — 100% free, 100% private, 100% local.**
---
✨ What This Is
This cheat sheet walks you through everything you need to set up Gemma 4 (Google's new open-weight AI model) running locally on your machine through Ollama, then connect it to Claude Code so you can code with AI assistance for $0/month — forever.
No API keys. No usage bills. No rate limits. No internet required.
---
🎯 Why You Want This
💰 Cost — Code without spending a penny
🔒 Privacy — Every line of code stays on your machine
⚡ Speed — No network round-trips, instant local inference
🧠 Performance — Good enough for 80% of everyday coding tasks
---
📋 Prerequisites
Before you start, make sure you have:
A Mac, Linux box, or Windows PC
Ollama v0.20.0 or later
At least 8GB RAM (e4b) or 20GB RAM (31b)
Claude Code installed
About 10 minutes
---
1️⃣ Install Ollama
Mac
```brew install ollama

Linux ```curl -fsSL ollama.com/install.sh ↗ | sh

Windows
Download from https://ollama.com/download
---
2️⃣ Start Server
```ollama serve

3️⃣ Pull Model ```ollama pull gemma4:31b

---
4️⃣ Test
```ollama run gemma4:31b

5️⃣ Connect Claude Code ```ollama launch claude

---
⚡ Quick Start
```brew install ollama && ollama serve &
ollama pull gemma4:31b
ANTHROPIC_BASE_URL=http://localhost:11434 ANTHROPIC_AUTH_TOKEN=ollama ANTHROPIC_API_KEY= ANTHROPIC_MODEL=gemma4:31b claude

🛠️ Commands ollama list ollama ps ollama stop gemma4:31b ollama rm gemma4:31b ollama --version

🎁 What You Get Free AI coding 100% local & private Offline usage No limits Apache 2.0 license

📚 Links ollama.com ↗ ollama.com/library/gemma4 ↗ docs.anthropic.com ↗

🦙 Ollama: ollama.com ↗

Gemma 4 + Ollama

1

Recursos

↑ voltar ao topo · ver no Telegram ↗