Pawan.KrdBeta
API Reference

Documentation

Everything you need to integrate Pawan.Krd into your application.

Introduction

Pawan.Krd is a unified, OpenAI-compatible API that gives developers and creators access to a broad catalog of AI models — chat & text LLMs, embeddings, rerankers, moderation, text-to-speech, speech-to-text, image, and video models — through a single base URL and a single API key.

Drop-in replace your OpenAI client by changing the baseURL and apiKey; everything else (request shapes, response objects, streaming over SSE, function/tool calling, vision inputs, etc.) follows the OpenAI specification.

Base URL

https://api.pawan.krd/v1

Model families

We don't just serve LLMs — Pawan.Krd routes requests to the right backend based on the model's declared types. Browse the full catalog on the Models page.

Chat & Text

Conversational LLMs and base/instruct text models with streaming, vision, tool calling, and reasoning support.

Embeddings

High-quality vector embeddings for search, RAG, clustering, and similarity.

Rerankers

Cross-encoder rerankers that score and reorder documents for a query.

Moderation

Safety classifiers that flag unsafe content across multiple categories.

Proofreading

Grammarly-style spelling, grammar, tone, pure-language and rewrite corrections.

Speech (TTS)

Text-to-speech with multiple voices, formats, and optional voice cloning.

Transcription (STT)

Speech-to-text and translation, with word-level timestamps and SRT/VTT output.

Image Models

Text-to-image generation, editing/inpainting, and upscaling.

Video Models

Text-to-video and image-to-video generation with configurable duration and FPS.

Why Pawan.Krd

  • OpenAI-compatible — works with the official OpenAI SDK and most third-party tools out of the box.
  • One API for many modalities: chat, text, embeddings, rerank, moderation, TTS, STT, image, and video.
  • Server-Sent Events streaming for chat, text, and audio.
  • Per-key access controls — IP allowlists, credit quotas, and request quotas.
  • Credit-based billing — predictable, no per-token surprises.
  • Path-based and body-based model selection (see Base URLs).

Where to next

Next

Authentication