Hello, world!

This page walks through the smallest possible Aspose.LLM for .NET program. It applies a license, creates an API instance from a preset, sends one message, and prints the response.

Prerequisites

  • Install the Aspose.LLM NuGet package.
  • Obtain a license (a free temporary license works for evaluation).
  • Enough RAM for the chosen model. Qwen25Preset needs roughly 8-12 GB: see System requirements.
  • Obtain the model file from its publisher’s chosen source after reviewing its license. Set ASPOSE_LLM_MODEL_FILE to its local path.
  • Internet access for native runtime binaries if they are not already cached.

Code

Step 1. Apply the license

Apply the license once, before constructing the API. Aspose.LLM does not run inference in evaluation mode.

using Aspose.LLM;

var license = new Aspose.LLM.License();
license.SetLicense("Aspose.LLM.lic");

Step 2. Create the API

Create an instance from a built-in preset. Qwen25Preset is a good general-purpose starting point.

using Aspose.LLM.Abstractions.Parameters.Presets;

var preset = new Qwen25Preset();
var modelPath = Environment.GetEnvironmentVariable("ASPOSE_LLM_MODEL_FILE");
if (string.IsNullOrWhiteSpace(modelPath) || !File.Exists(modelPath))
    throw new InvalidOperationException("Provide a local Qwen 2.5 7B Instruct GGUF via ASPOSE_LLM_MODEL_FILE. Preset source: https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-GGUF; model license: https://www.apache.org/licenses/LICENSE-2.0");
preset.BaseModelSourceParameters.ModelFilePath = modelPath;
using var api = AsposeLLMApi.Create(preset);

The using pattern disposes the API automatically when the block exits. Disposal unloads the model and releases the single-instance guard.

Step 3. Send a message

If no chat session exists yet, SendMessageAsync creates one implicitly.

string response = await api.SendMessageAsync("Say hello in one short sentence.");
Console.WriteLine(response);
// Output: Hello! Nice to meet you.

Full example

using Aspose.LLM;
using Aspose.LLM.Abstractions.Parameters.Presets;

var license = new Aspose.LLM.License();
license.SetLicense("Aspose.LLM.lic");

var preset = new Qwen25Preset();
var modelPath = Environment.GetEnvironmentVariable("ASPOSE_LLM_MODEL_FILE");
if (string.IsNullOrWhiteSpace(modelPath) || !File.Exists(modelPath))
    throw new InvalidOperationException("Provide a local Qwen 2.5 7B Instruct GGUF via ASPOSE_LLM_MODEL_FILE. Preset source: https://huggingface.co/bartowski/Qwen2.5-7B-Instruct-GGUF; model license: https://www.apache.org/licenses/LICENSE-2.0");
preset.BaseModelSourceParameters.ModelFilePath = modelPath;
using var api = AsposeLLMApi.Create(preset);

string response = await api.SendMessageAsync("Say hello in one short sentence.");
Console.WriteLine(response);

Save this as Program.cs in a new .NET 8 or .NET 10 project, ensure Aspose.LLM.lic is next to the executable, and run:

dotnet run

Troubleshooting

  • Not licensed for this method: the license has not been applied. Call SetLicense before AsposeLLMApi.Create. See Licensing.
  • First Create takes a long time: the native runtime may need downloading, and model loading can take time. Check the local model path and monitor progress via ILogger (pass one to AsposeLLMApi.Create(preset, logger)).
  • Only one AsposeLLMApi instance can be created at a time: a previous instance has not been disposed yet. Keep one instance for the process lifetime, or dispose the old one first.

What’s next

How to run examples