> ## 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.

# Why Forii

> How Forii compares to other inference providers for Indian developers

Forii is built specifically for India — frontier models at lower cost, on Indian infrastructure, with full data sovereignty and INR pricing.

## Comparison

| Dimension                 | Forii                                         | OpenAI              | Fireworks           | Sarvam             |
| ------------------------- | --------------------------------------------- | ------------------- | ------------------- | ------------------ |
| **India data residency**  | Yes — Delhi NCR                               | No (US/EU)          | No (US)             | Yes — India        |
| **Data jurisdiction**     | Indian law                                    | US (CLOUD Act)      | US (CLOUD Act)      | Indian law         |
| **Pricing currency**      | INR (₹)                                       | USD (\$)            | USD (\$)            | INR (₹)            |
| **OpenAI-compatible API** | Yes                                           | Yes                 | Yes                 | Partial            |
| **Payment methods**       | UPI, netbanking, cards, wallets (coming soon) | Card only           | Card only           | UPI, cards         |
| **Hindi-strong models**   | Evaluated on MMMU-Hindi                       | No Hindi evaluation | No Hindi evaluation | 22 Indic languages |
| **Chat completions**      | Yes                                           | Yes                 | Yes                 | Yes                |
| **Embeddings**            | Yes                                           | Yes                 | Yes                 | No                 |
| **Streaming**             | Yes                                           | Yes                 | Yes                 | Yes                |
| **Structured outputs**    | Yes                                           | Yes                 | Yes                 | Limited            |
| **Function calling**      | Yes                                           | Yes                 | Yes                 | No                 |
| **STT / TTS**             | Coming soon                                   | Yes                 | No                  | Yes                |
| **Vision**                | Coming soon                                   | Yes                 | Yes                 | Limited            |
| **Fine-tuning**           | Coming soon                                   | Yes                 | Yes                 | No                 |
| **Batch inference**       | Coming soon                                   | Yes                 | Yes                 | No                 |
| **Latency from India**    | \~20-50ms TTFT                                | \~200-400ms TTFT    | \~200-400ms TTFT    | \~20-50ms TTFT     |

## Where Forii wins

**Frontier models, lower cost.** Run DeepSeek-V3, LLaMA-4-Scout, Gemma-3, and Qwen3 at 30% below self-deployment cost. Continuous batching, INT4/AWQ quantization, and prompt caching compound the savings.

**Data sovereignty.** Indian data centers, Indian jurisdiction. No data routed through US servers, no US CLOUD Act exposure, no foreign subpoenas. Verify with the `x-forii-region` header on every request.

**INR pricing.** Pay in rupees. No FX conversion. Paid tiers with UPI/card payments and GST-compliant invoices are on the roadmap; today the Free Plan needs no payment at all.

**OpenAI-compatible.** Change `base_url` and `api_key` — that's it. Every existing tutorial, framework, and SDK works. No vendor lock-in.

**Hindi quality verified.** Every model is evaluated on MMMU-Hindi before deployment. If quantization degrades Hindi quality, the model is rejected.

## Where others win

**OpenAI** has more models (GPT-4o, DALL-E, Whisper) and the largest ecosystem. If you need the absolute best model quality regardless of cost, OpenAI is still the benchmark.

**Fireworks** has fine-tuning (SFT, DPO, RFT) available today, deployments, and speculative decoding. If you need fine-tuning right now, Fireworks is ahead.

**Sarvam** has production-grade STT, TTS, and translation for 22 Indic languages today. If voice AI is your primary use case, Sarvam is the leader.

**Baseten** has the most production-grade deployment story (config-only Truss, TensorRT, SSH debug). If you need custom model deployment at scale, Baseten has deeper infra tooling.

Forii's edge is frontier models at lower cost, with data sovereignty on Indian infrastructure and INR pricing. STT, TTS, vision, and fine-tuning are coming soon to close the gaps.

## Related

* [Overview](/docs/concepts/overview) — How Forii works
* [Models](/docs/concepts/models) — Available models and pricing
* [Quick Start](/docs/quickstart) — Make your first API call