Deploying Ollama on Windows and Linux and Integrating with SpringBoot

Ollama is an open‑source framework designed for easy local deployment and execution of large language models (LLMs).

Download Ollama : The official download page (https://ollama.com/download) provides binaries for macOS, Linux, and Windows. The GitHub releases page (https://github.com/ollama/ollama/releases) can also be used. For example, on Windows you can fetch the installer with:

wget https://github.com/ollama/ollama/releases/download/v0.5.8-rc10/OllamaSetup.exe

After downloading, run the installer (default to C:\, requiring at least 10 GB free). Verify the installation by opening a command prompt and typing ollama; a usage message indicates success. Close any running Ollama process before proceeding.

Configure environment variables to control model storage and network settings: OLLAMA_MODELS – path where models are stored, e.g.,

D:\ollama\models

OLLAMA_HOST

– optional custom host, e.g., 0.0.0.0:8080 (useful when a UI expects port 8080) OLLAMA_ORIGINS – optional CORS setting, e.g., * Select and download a DeepSeek model . Visit the model library (https://ollama.com/library/deepseek-r1) and choose a size that matches your hardware. For a 1.5 B parameter model, run: ollama run deepseek-r1:1.5b The command pulls the model and makes it ready for inference.

Linux deployment can be performed with a single script:

curl -fsSL https://ollama.com/install.sh | sh
sudo systemctl enable ollama
sudo systemctl status ollama
sudo systemctl start ollama
ollama run deepseek-r1:1.5b

API integration (SpringBoot example) :

The Ollama server exposes a comprehensive REST API (see https://github.com/ollama/ollama/blob/main/docs/api.md). The following utility class sends a generation request to the /api/generate endpoint:

package com.example.springtestdemo.ai.util;

import com.alibaba.fastjson.JSONObject;
import lombok.extern.slf4j.Slf4j;

@Slf4j
public class OllamaUtil {
    private static String DOMAIN = "http://localhost:11434/api/generate";

    public static String chatDeepSeek(String model, String question) {
        String url = DOMAIN;
        JSONObject body = new JSONObject();
        body.put("model", model);
        body.put("prompt", question);
        body.put("stream", false);
        String result = CommonUtil.postJson(url, body.toJSONString());
        log.info("【ollama-请求】 结果 {}", result);
        try {
            JSONObject resJson = JSONObject.parseObject(result);
            String response = resJson.getString("response");
            log.info("【ollama-请求】 结果 {}", response);
            return response;
        } catch (Exception e) {
            log.error("【ollama-请求】异常", e);
        }
        return "ok";
    }
}

The corresponding controller exposes a POST endpoint:

package com.example.springtestdemo.ai.controller;

import com.example.springtestdemo.ai.QueryParam;
import com.example.springtestdemo.ai.util.OllamaUtil;
import org.springframework.web.bind.annotation.PostMapping;
import org.springframework.web.bind.annotation.RestController;

@RestController
public class AiController {
    @PostMapping("/deepSeek")
    public String deepSeek(QueryParam query) {
        String res = OllamaUtil.chatDeepSeek("deepseek-r1:1.5b", query.getQuestion());
        return res;
    }
}

For a graphical UI, the article suggests using an Edge browser extension (https://www.crxsoso.com/webstore/detail/jfgfiigpkhlkbnfnbobbkinehhfdhndo) that supports image upload and other features.

Overall, the guide provides step‑by‑step instructions for installing Ollama, configuring it, pulling a DeepSeek model, and calling the service from Java code.