> ## Documentation Index
> Fetch the complete documentation index at: https://docs.forii.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> How Forii works — India's Sovereign Inference Platform

Forii is an OpenAI-compatible inference API that lets you run any frontier model on Indian infrastructure — independent of US cloud providers. 30% lower cost, full data sovereignty, INR pricing.

## How it works

<img src="https://mintcdn.com/forii-docs/cNsW6DAG7qkXCpMH/docs/images/overview-flow.svg?fit=max&auto=format&n=cNsW6DAG7qkXCpMH&q=85&s=28346289ed481119e736968e474c02c0" alt="Forii request flow: your app → API gateway → inference → response pipeline" width="800" height="500" data-path="docs/images/overview-flow.svg" />

## Available now

| Capability       | Endpoint                              | Description                                            |
| ---------------- | ------------------------------------- | ------------------------------------------------------ |
| Chat completions | `POST /inference/v1/chat/completions` | Basic, streaming, structured outputs, function calling |
| Embeddings       | `POST /inference/v1/embeddings`       | Semantic search over Hindi and English documents       |
| Models           | `GET /inference/v1/models`            | List available models                                  |
| Account & usage  | `GET/POST /v1/accounts/{id}/...`      | API keys, balance, usage queries                       |

## Coming soon

| Capability      | Description                                     |
| --------------- | ----------------------------------------------- |
| Speech-to-text  | 22 Indic languages, 8kHz telephony              |
| Text-to-speech  | 30+ voices across Hindi, Tamil, Telugu, Bengali |
| Vision          | Image understanding — Aadhaar, PAN, invoice OCR |
| Reranking       | Cross-encoder reranking for RAG                 |
| Batch inference | Asynchronous large-scale processing             |
| Fine-tuning     | SFT on Indic-language data                      |
| Deployments     | Dedicated GPU with autoscaling                  |

## OpenAI compatibility

Forii matches the OpenAI API format exactly. Every existing tutorial, framework, and SDK works by changing two lines:

```python theme={null}
from openai import OpenAI

client = OpenAI(
    base_url="https://api.forii.in/inference/v1",  # ← change this
    api_key=os.environ["FORII_API_KEY"],            # ← and this
)

# Everything else is identical to OpenAI
response = client.chat.completions.create(
    model="forii/deepseek-v3",
    messages=[{"role": "user", "content": "Hello"}],
)
```

### Framework compatibility

| Framework             | Works? | How                                 |
| --------------------- | ------ | ----------------------------------- |
| OpenAI Python SDK     | Yes    | `base_url` + `api_key`              |
| OpenAI JavaScript SDK | Yes    | `baseURL` + `apiKey`                |
| LangChain             | Yes    | `ChatOpenAI` with Forii `base_url`  |
| LlamaIndex            | Yes    | `OpenAI` with Forii `base_url`      |
| Vercel AI SDK         | Yes    | `createOpenAI` with Forii `baseURL` |
| LiteLLM               | Yes    | Add Forii as a custom provider      |

## India-first design

* **Frontier models, lower cost** — DeepSeek-V3, LLaMA-4-Scout, Gemma-3, Qwen3 at 30% below self-deployment cost.
* **Data sovereignty** — Indian data centers, Indian jurisdiction. No US routing, no CLOUD Act exposure, no foreign subpoenas.
* **Indian data centers** — Requests served from Delhi NCR. \~20-50ms TTFT vs 200-400ms from US providers.
* **INR pricing** — All costs in rupees. No USD invoices, no FX conversion.
* **Free Plan today, paid tiers coming** — Start free with no card. UPI payments, GST invoices, and higher limits ship with paid tiers.

## Next steps

* [Models](/docs/concepts/models) — Available models and pricing
* [Authentication](/docs/concepts/authentication) — Set up your API key
* [Chat Completions](/docs/api-reference/chat-completions) — Full API reference