--- title: "Expose the Ollama API via a Public API Implementation in .NET Core" description: "Ollama enables developers to run Large Language Models (LLMs) locally. However, many real-world applications require external access to these models through a p" author: "Anubhav Sharma" published: 2026-06-09 updated: 2026-06-10 canonical: https://answers.mindstick.com/blog/384/expose-the-ollama-api-via-a-public-api-implementation-in-dot-net-core category: "programming" tags: ["ollama", "programming language", "artificial intelligence"] reading_time: 4 minutes --- # Expose the Ollama API via a Public API Implementation in .NET Core [Ollama](https://answers.mindstick.com/blog/350/understanding-ollama-api) enables developers to run [Large Language Models (LLMs)](https://www.mindstick.com/interview/34471/what-is-a-large-language-model-llm) locally. However, many real-world applications require external access to these models through a public API. ![Expose the Ollama API via a Public API Implementation in .NET Core](https://answers.mindstick.com/blogs/8995c16a-47f5-429a-ae6b-45e514a0fcb1/images/3534b131-768e-4143-8952-c85549099ad8.png) ## Prerequisites Before starting, ensure that: - .NET 8 SDK is installed. - Ollama is installed and running. - At least one model is downloaded. Check available models: ```plaintext ollama list ``` Example: ```plaintext llama3.2 mistral gemma3 deepseek-r1 ``` Start Ollama: ```plaintext ollama serve ``` By default, Ollama runs on: ```plaintext http://localhost:11434 ``` ## Step 1: Create a .NET Core Web API Create a new project: ```plaintext dotnet new webapi -n OllamaPublicApi cd OllamaPublicApi ``` Run the project: ```plaintext dotnet run ``` ## Step 2: Create Request Models Create a folder named **Models**. ### Models/ChatRequest.cs ```cs namespace OllamaPublicApi.Models { public class ChatRequest { // AI model name public string Model { get; set; } // User prompt public string Prompt { get; set; } } } ``` ## Step 3: Register HttpClient Open **Program.cs**. ```cs // Create builder var builder = WebApplication.CreateBuilder(args); // Register controllers builder.Services.AddControllers(); // Register HttpClient for API calls builder.Services.AddHttpClient(); // Enable Swagger builder.Services.AddEndpointsApiExplorer(); builder.Services.AddSwaggerGen(); var app = builder.Build(); // Enable Swagger UI app.UseSwagger(); app.UseSwaggerUI(); // Map controllers app.MapControllers(); // Run application app.Run(); ``` ## Step 4: Create Ollama Controller Create a folder named **Controllers**. ### Controllers/OllamaController.cs ```cs using Microsoft.AspNetCore.Mvc; using OllamaPublicApi.Models; using System.Text; using System.Text.Json; namespace OllamaPublicApi.Controllers { [ApiController] [Route("api/[controller]")] public class OllamaController : ControllerBase { // HttpClient factory instance private readonly IHttpClientFactory _httpClientFactory; // Constructor injection public OllamaController(IHttpClientFactory httpClientFactory) { _httpClientFactory = httpClientFactory; } [HttpPost("chat")] public async Task Chat(ChatRequest request) { // Create HttpClient var client = _httpClientFactory.CreateClient(); // Build Ollama request payload var payload = new { model = request.Model, prompt = request.Prompt, stream = false }; // Convert object into JSON string var json = JsonSerializer.Serialize(payload); // Create HTTP content var content = new StringContent( json, Encoding.UTF8, "application/json"); // Send request to Ollama API var response = await client.PostAsync( "http://localhost:11434/api/generate", content); // Read response body var result = await response.Content.ReadAsStringAsync(); // Return response to client return Content(result, "application/json"); } } } ``` ## Step 5: Test Dynamic Model Selection ### Request Using Llama ```plaintext POST /api/ollama/chat ``` ```plaintext { "model": "llama3.2", "prompt": "Explain machine learning." } ``` ### Request Using Mistral ```plaintext { "model": "mistral", "prompt": "Explain machine learning." } ``` ### Request Using DeepSeek ```plaintext { "model": "deepseek-r1", "prompt": "Explain machine learning." } ``` The API automatically forwards the selected model to Ollama. ## Step 6: Create Model Listing Endpoint Allow clients to discover available models. Add the following method inside **OllamaController**. ```cs [HttpGet("models")] public async Task GetModels() { // Create HttpClient var client = _httpClientFactory.CreateClient(); // Call Ollama tags endpoint var response = await client.GetAsync( "http://localhost:11434/api/tags"); // Read response var result = await response.Content.ReadAsStringAsync(); // Return model list return Content(result, "application/json"); } ``` Test: ```plaintext GET /api/ollama/models ``` Example Response: ```plaintext { "models": [ { "name": "llama3.2" }, { "name": "mistral" } ] } ``` ## Step 7: Add API Key Security Add the following code to the Chat method. ```cs // Read API Key from request header var apiKey = Request.Headers["X-API-KEY"].ToString(); // Validate API Key if (apiKey != "my-secret-key") { // Return unauthorized response return Unauthorized("Invalid API Key"); } ``` Client Request: ```plaintext POST /api/ollama/chat ``` Headers: ```plaintext X-API-KEY: my-secret-key ``` This prevents unauthorized access. ## Step 8: Expose API Publicly ## Option 1: Ngrok Install Ngrok and run: ```plaintext ngrok http 5000 ``` Example: ```plaintext https://abc123.ngrok-free.app ``` Public Endpoint: ```plaintext https://abc123.ngrok-free.app/api/ollama/chat ``` ## Option 2: Deploy to Cloud Deploy your [application](https://www.mindstick.com/blog/59/xaml-extensible-application-markup-language) to: - [AWS EC2](https://aws.amazon.com/ec2/customers/) - Azure Virtual Machine - [Google Cloud VM](https://cloud.google.com/learn/what-is-a-virtual-machine) - DigitalOcean Install Ollama on the server: ```plaintext ollama serve ``` Publish the .NET application: ```plaintext dotnet publish -c Release ``` Run: ```plaintext dotnet OllamaPublicApi.dll ``` Configure firewall and reverse proxy settings. ## Recommended Models | Model | Purpose | | --- | --- | | llama3.2 | General AI assistant | | mistral | Fast responses | | gemma3 | Lightweight applications | | deepseek-r1 | Reasoning and coding | | qwen3 | Multilingual tasks | | codellama | Code generation | ## Conclusion Using [ASP.NET Core and Ollama](https://answers.mindstick.com/blog/375/building-an-enterprise-ai-search-engine-with-ollama), you can quickly build a secure public AI API. The [implementation](https://www.mindstick.com/interview/733/what-is-implementation-and-interface-inheritance) allows clients to dynamically choose models, making it suitable for chatbots, AI assistants, content generation platforms, and [enterprise AI solutions](https://www.mindstick.com/articles/337935/what-is-amazon-q-developer-ai-chatbot-for-companies). By adding model discovery, API key [authentication](https://www.mindstick.com/blog/177/authentication-and-authorization-in-asp-dot-net), and public [deployment](https://www.mindstick.com/articles/332756/exploring-the-benefits-of-continuous-integration-and-deployment), you can create a production-ready AI service powered by locally hosted LLMs. --- Original Source: https://answers.mindstick.com/blog/384/expose-the-ollama-api-via-a-public-api-implementation-in-dot-net-core Copyright © MindStick Software Pvt. Ltd. This Markdown version is provided for developers, AI systems, and offline reading.