Ollama: 137K+ Stars — 一条命令本地运行大模型，2026 完整配置指南

# 使用 Homebrew（推荐）
brew install ollama

# 或从 ollama.com/download 下载原生应用

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

ollama --version
# ollama version 0.6.7

# 启动守护进程（如果尚未运行）
ollama serve

# 拉取并运行你的第一个模型
ollama run llama3.2:8b

# 使用 Docker 运行 Open WebUI
docker run -d -p 3000:8080 \
  --add-host=host.docker.internal:host-gateway \
  -v open-webui:/app/backend/data \
  --name open-webui \
  --restart always \
  ghcr.io/open-webui/open-webui:main

# 安装
pip install langchain-ollama

# 聊天模型
from langchain_ollama import ChatOllama

llm = ChatOllama(
    model="llama3.2:8b",
    temperature=0.7,
    base_url="http://localhost:11434"
)

response = llm.invoke("用一段话解释量子计算。")
print(response.content)

# 嵌入向量
from langchain_ollama import OllamaEmbeddings

embeddings = OllamaEmbeddings(model="nomic-embed-text")
vector = embeddings.embed_query("Hello world")
# 返回 768 维浮点向量

{
  "models": [
    {
      "title": "Llama 3.2",
      "provider": "ollama",
      "model": "llama3.2:8b",
      "apiBase": "http://localhost:11434"
    }
  ],
  "tabAutocompleteModel": {
    "title": "CodeQwen",
    "provider": "ollama",
    "model": "codeqwen:7b-code"
  }
}

# 生成文本
curl http://localhost:11434/api/generate -d '{
  "model": "llama3.2:8b",
  "prompt": "天空为什么是蓝色的？",
  "stream": false
}'

# 聊天补全（兼容 OpenAI）
curl http://localhost:11434/v1/chat/completions -H "Content-Type: application/json" -d '{
  "model": "llama3.2:8b",
  "messages": [{"role": "user", "content": "你好！"}],
  "temperature": 0.7
}'

# 生成嵌入向量
curl http://localhost:11434/api/embed -d '{
  "model": "nomic-embed-text",
  "input": ["天空是蓝色的", "草地是绿色的"]
}'

# docker-compose.yml
version: "3.8"

services:
  ollama:
    image: ollama/ollama:0.6.7
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama_data:/root/.ollama
    environment:
      - OLLAMA_KEEP_ALIVE=24h
      - OLLAMA_NUM_PARALLEL=4
      - OLLAMA_MAX_LOADED_MODELS=2
    restart: unless-stopped
    # NVIDIA GPU 支持
    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"
    environment:
      - OLLAMA_BASE_URL=http://ollama:11434
    volumes:
      - openwebui_data:/app/backend/data
    depends_on:
      - ollama
    restart: unless-stopped

volumes:
  ollama_data:
  openwebui_data:

# 安装 NVIDIA Container Toolkit
curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey \
  | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg

curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list \
  | 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-get update
sudo apt-get install -y nvidia-container-toolkit
sudo nvidia-ctk runtime configure --runtime=docker
sudo systemctl restart docker

services:
  ollama:
    image: ollama/ollama:rocm
    devices:
      - /dev/kfd
      - /dev/dri
    group_add:
      - video
    environment:
      - HSA_OVERRIDE_GFX_VERSION=11.0.0

services:
  ollama:
    image: ollama/ollama:0.6.7
    environment:
      - OLLAMA_NUM_PARALLEL=4      # 4 个并发请求
      - OLLAMA_MAX_LOADED_MODELS=2  # VRAM 中保持 2 个模型
      - OLLAMA_KEEP_ALIVE=30m      # 空闲 30 分钟后卸载
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

# Modelfile
FROM llama3.2:8b

# 系统提示词定义人格
SYSTEM """你是一位资深软件工程师。回答要简洁、实用，
并且始终包含可运行的代码示例。"""

# 参数调优
PARAMETER temperature 0.3
PARAMETER num_ctx 16384
PARAMETER top_p 0.9
PARAMETER repeat_penalty 1.1
PARAMETER stop "<|eot_id|>"

# 自定义模板（可选 —— 省略则继承基础模型）
TEMPLATE """{{ if .System }}<|start_header_id|>system<|end_header_id|>

{{ .System }}<|eot_id|>{{ end }}{{ if .Prompt }}<|start_header_id|>user<|end_header_id|>

{{ .Prompt }}<|eot_id|>{{ end }}<|start_header_id|>assistant<|end_header_id|>

{{ .Response }}<|eot_id|>"""

# 创建自定义模型
ollama create senior-dev -f Modelfile

# 运行
ollama run senior-dev

# 查看生效的 Modelfile
ollama show senior-dev --modelfile

# Modelfile.code-review
FROM codellama:7b-code

SYSTEM """你是一个代码审查助手。分析提供的代码，检查：
1. Bug 和逻辑错误
2. 安全漏洞（SQL 注入、XSS、缓冲区溢出）
3. 性能问题（N+1 查询、不必要的内存分配）
4. 风格和可读性

按以下格式回复：
- [CRITICAL] 用于 Bug/安全问题
- [WARN] 用于性能问题
- [INFO] 用于风格建议

始终为 [CRITICAL] 和 [WARN] 项提供修复建议。"""

PARAMETER temperature 0.1
PARAMETER num_ctx 8192
PARAMETER num_predict 2048

ollama create code-reviewer -f Modelfile.code-review

# Modelfile.local
FROM ./my-fine-tuned-model-q4_k_m.gguf

PARAMETER temperature 0.7
PARAMETER num_ctx 4096

SYSTEM "你是一个专注于医学术语的有用助手。"

ollama create med-assistant -f Modelfile.local

# 显示模型详情和 Modelfile
ollama show llama3.2:8b --modelfile

# 仅显示参数
ollama show llama3.2:8b --parameters

# 显示系统提示词
ollama show llama3.2:8b --system

# 列出所有本地模型
ollama list

# 显示运行中的模型
ollama ps

# 核心设置
OLLAMA_HOST=0.0.0.0:11434          # 绑定到所有接口
OLLAMA_KEEP_ALIVE=24h               # 模型保持加载 24 小时
OLLAMA_NUM_PARALLEL=4               # 最大并发请求数
OLLAMA_MAX_LOADED_MODELS=2          # VRAM 中同时加载的最大模型数
OLLAMA_FLASH_ATTENTION=1            # 启用 Flash Attention（更快推理）

# 性能调优
OLLAMA_GPU_OVERHEAD=200MB           # 预留 VRAM 余量
OLLAMA_DEBUG=1                      # 详细日志

server {
    listen 443 ssl http2;
    server_name ollama.yourdomain.com;

    ssl_certificate /etc/letsencrypt/live/ollama.yourdomain.com/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/ollama.yourdomain.com/privkey.pem;

    location / {
        proxy_pass http://localhost:11434;
        proxy_http_version 1.1;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
        
        # 流式传输的 WebSocket 支持
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
        
        # 长时间推理的超时设置
        proxy_read_timeout 600s;
        proxy_send_timeout 600s;
    }
}

# ollama-auth-proxy.py（Flask 示例）
from flask import Flask, request, Response
import requests

app = Flask(__name__)
OLLAMA_URL = "http://localhost:11434"
VALID_KEYS = {"sk-your-api-key-here"}

@app.route('/', defaults={'path': ''}, methods=['GET', 'POST', 'PUT', 'DELETE'])
@app.route('/<path:path>', methods=['GET', 'POST', 'PUT', 'DELETE'])
def proxy(path):
    api_key = request.headers.get('Authorization', '').replace('Bearer ', '')
    if api_key not in VALID_KEYS:
        return {"error": "无效的 API 密钥"}, 401
    
    resp = requests.request(
        method=request.method,
        url=f"{OLLAMA_URL}/{path}",
        headers={k: v for k, v in request.headers if k != 'Host'},
        data=request.get_data(),
        stream=True
    )
    return Response(resp.iter_content(chunk_size=1024), status=resp.status_code,
                   content_type=resp.headers.get('Content-Type'))

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=11435)

# 列出运行中的模型及其内存使用情况
curl http://localhost:11434/api/ps

# /etc/systemd/system/ollama.service
[Unit]
Description=Ollama LLM 服务
After=network-online.target

[Service]
ExecStart=/usr/local/bin/ollama serve
User=ollama
Group=ollama
Restart=always
RestartSec=3
Environment="OLLAMA_HOST=0.0.0.0:11434"
Environment="OLLAMA_NUM_PARALLEL=4"
Environment="OLLAMA_KEEP_ALIVE=24h"

[Install]
WantedBy=default.target

sudo systemctl daemon-reload
sudo systemctl enable ollama
sudo systemctl start ollama