OpenAI-Compatible Model APIs with Prompt Caching

Use one Lumino API key for hosted chat, coding, reasoning, video, and speech models. Keep familiar OpenAI client patterns, pay in INR, and inspect request-level token usage and cost.

Change the base URL, not your application

const client = new OpenAI({ apiKey: process.env.LUMINO_API_KEY, baseURL: "https://api.luminoai.co.in/api/v1" });

Call POST /api/v1/chat/completions with model IDs from the live catalog. Streaming chat completions use the same API key.

Read the Lumino API docs Browse the live model catalog

Prompt caching with visible usage

Supported models report cached input tokens so repeated system prompts, document context, and agent instructions can be measured. In one repeated-prefix test, the first request used 3,066 prompt tokens with 0 cached and cost 51 paise; the repeated request reported 3,040 cached tokens and cost 12 paise.

See the prompt-caching benchmark

Compare Lumino, Fireworks AI, and fal by workload

Provider prices and model availability change. Compare the same model, billing unit, cached-input rate, output rate, and payment friction before switching. Lumino is designed for Indian teams that value INR billing, one API key across model families, and cached-token visibility.

Production usage

Lumino has served more than 45 million tokens across production model API workloads.

Create a Lumino API key