.NET AI API client (India)

A .NET service can call unoblox with nothing more than HttpClient, since the endpoint mirrors the same OpenAI-compatible chat-completions shape used across the industry. One base URL and one key give a C# application access to GPT, Claude, DeepSeek, Qwen, and the rest of the catalog, billed per token in rupees on a single monthly GST invoice.

Why plain HttpClient is enough

Community OpenAI-compatible client packages for .NET generally let you override the base address, so they work against unoblox unchanged. If you would rather avoid an extra dependency, the request and response are plain JSON over HTTPS, which HttpClient and System.Text.Json handle without any extra package.

A minimal request

using System.Net.Http;
using System.Net.Http.Headers;
using System.Text;

var client = new HttpClient();
client.DefaultRequestHeaders.Authorization =
    new AuthenticationHeaderValue("Bearer", "ub-gw-xxxxxxxxxxxxxxxx");

var payload = "{\"model\":\"openai/gpt-4.1\",\"messages\":[{\"role\":\"user\",\"content\":\"Summarise this support thread in two sentences.\"}]}";

var response = await client.PostAsync(
    "https://api.unoblox.ai/v1/chat/completions",
    new StringContent(payload, Encoding.UTF8, "application/json"));

Console.WriteLine(await response.Content.ReadAsStringAsync());

Building the JSON payload with System.Text.Json.JsonSerializer instead of a hand-written string is the more maintainable option in a real project; the raw string above just keeps the example to one file.

Choosing a model from a .NET service

Use case	Model	₹ per 1M tokens (input / output)
Background worker jobs	DeepSeek V4 Flash	₹9.07 / ₹18.14
ASP.NET request-time feature	Llama 4 Maverick	₹20.16 / ₹80.64
Customer-facing assistant	GPT-5	₹126 / ₹1008
Highest-quality output	Claude Opus	₹504 / ₹2520

Anything outside this table, see live ₹ pricing on /models before wiring a new model id into a controller.

Fitting this into an ASP.NET Core project

Register a named HttpClient for the unoblox base address through IHttpClientFactory, keep the key in configuration, such as user secrets locally and an environment-backed configuration provider in production, rather than in source, and surface the model id and token usage from each response in your existing logging or telemetry pipeline so cost is visible per endpoint.

Frequently asked questions

Do I need a NuGet package to call unoblox? No — HttpClient and System.Text.Json, both part of the base class library, are enough; a community OpenAI-compatible NuGet package works too if you prefer a typed client.

Does this work the same from a Blazor Server or MAUI app? Yes, since the call is a standard outbound HTTPS request, it works from any .NET application type that permits network calls, not just ASP.NET Core.

Is data processed inside India? Only for the small unoblox-hosted models in the catalog; external models like GPT and Claude run on their own infrastructure, with unoblox adding rupee billing and a GST invoice rather than India data residency for those calls.

How should I handle the API key in production? Store it in your platform's configuration or secrets store, for example Azure App Configuration or an environment-backed provider, and inject it via IOptions, the same way you would handle any third-party API credential.

Can I stream a response in .NET? Yes, read the response stream incrementally with HttpCompletionOption.ResponseHeadersRead instead of waiting for the full body, the standard pattern for consuming server-sent events in .NET.

Is there a free model to validate the integration first? Yes, Qwen3 1.7B is free (₹0), enough to confirm your HttpClient setup and JSON parsing before switching to a paid model.

Get started in rupees → https://unoblox.ai/sign-in