推送模型 (POST /api/push)

推送模型接口用于将本地模型上传到 Ollama 模型库或私有仓库。

基本用法

curl http://localhost:11434/api/push -d '{
  "name": "myusername/mymodel"
}'

响应（流式）：

{"status":"retrieving manifest"}
{"status":"pushing manifest"}
{"status":"pushing layer","digest":"abc123...","total":4661224676,"completed":0}
{"status":"pushing layer","digest":"abc123...","total":4661224676,"completed":1000000000}
...
{"status":"success"}

请求参数

模型命名规范

推送的模型需要包含用户名前缀：

username/modelname
username/modelname:tag

例如：

myuser/my-llama
myuser/my-llama:v1.0

代码示例

Python

import requests

def push_model(model_name):
    response = requests.post(
        "http://localhost:11434/api/push",
        json={"name": model_name},
        stream=True
    )
    
    for line in response.iter_lines():
        if line:
            import json
            data = json.loads(line)
            status = data.get("status", "")
            
            if "pushing layer" in status:
                total = data.get("total", 0)
                completed = data.get("completed", 0)
                if total > 0:
                    percent = (completed / total) * 100
                    print(f"\r上传进度: {percent:.1f}%", end="", flush=True)
            else:
                print(status)
    
    print("\n推送完成")

push_model("myuser/my-llama")

JavaScript

async function pushModel(modelName) {
    const response = await fetch('http://localhost:11434/api/push', {
        method: 'POST',
        headers: { 'Content-Type': 'application/json' },
        body: JSON.stringify({ name: modelName })
    });
    
    const reader = response.body.getReader();
    const decoder = new TextDecoder();
    
    while (true) {
        const { done, value } = await reader.read();
        if (done) break;
        
        const lines = decoder.decode(value).split('\n').filter(Boolean);
        for (const line of lines) {
            const data = JSON.parse(line);
            console.log(data.status);
        }
    }
    
    console.log('推送完成');
}

await pushModel('myuser/my-llama');

Go

package main

import (
    "bytes"
    "encoding/json"
    "fmt"
    "io"
    "net/http"
)

type PushRequest struct {
    Name string `json:"name"`
}

func pushModel(modelName string) error {
    req := PushRequest{Name: modelName}
    body, _ := json.Marshal(req)
    
    resp, err := http.Post(
        "http://localhost:11434/api/push",
        "application/json",
        bytes.NewReader(body),
    )
    if err != nil {
        return err
    }
    defer resp.Body.Close()
    
    decoder := json.NewDecoder(resp.Body)
    for {
        var data map[string]interface{}
        if err := decoder.Decode(&data); err != nil {
            if err == io.EOF {
                break
            }
            continue
        }
        fmt.Println(data["status"])
    }
    
    return nil
}

func main() {
    pushModel("myuser/my-llama")
}

推送到私有仓库

def push_to_private(model_name, registry="my-registry.com"):
    full_name = f"{registry}/{model_name}"
    push_model(full_name)

push_to_private("my-llama", "registry.example.com")

认证

推送模型需要先登录：

# 登录 Ollama
ollama login

# 或设置环境变量
export OLLAMA_USERNAME=myuser
export OLLAMA_PASSWORD=mypassword

注意事项

1. 需要账号：必须有 Ollama 账号
2. 模型命名：必须包含用户名前缀
3. 网络要求：需要能访问 ollama.ai
4. 上传时间：大模型上传需要时间