nepal·ai·gateway

API live · checking seats

One key. The models that are actually online.

Free plan routes to the stealth model diwakar, at no cost. Pro adds providers as each one is verified to work — so every slug you see is one you can actually call. Built to be reached from Nepal.

Make your first request See how it works

Works with the OpenAI SDKs and tools you already use.

// model catalog

What your key can call — live.

The live lineup on one OpenAI-compatible endpoint — same base URL for every model. Your account is the source of truth for what your key can call.

NEW · STEALTH POOL Zero data training

diwakar Multimodal

diwakar — a private, stealth-routed multimodal model, available for a limited time. The upstream does not train on your data.

diwakar Included on Free & Pro tiers
Context window 200K tokens
Capabilities Vision + Text
Privacy Policy Zero retention
SDK Protocol OpenAI / Anthropic
FREE TRIAL TIER

Instant sandbox.

Get started in seconds with Google Sign-in. Includes 5M daily tokens for diwakar with 10 RPM — zero card required.

diwakar5M/day10 RPM
PRO DEVELOPER

Production gateway pool.

50M monthly tokens with automatic multi-provider failover. Live verified catalog — only models your key can actually call.

50M/mo60 RPM/v1/models

Your account is the source of truth for enabled models — or call GET /v1/models before you choose.

// the honest scoreboard

Everything on this page is verifiable.

No invented numbers. The gateway publishes what it can actually serve — check any of it yourself right now.

Live API health →
/v1/models See exactly what's online →
100% Slugs backed by a working provider
Instant Free trial key — no card, no wait

Every claim on this page points to something you can curl in under a minute.

— The integrity promise

// why this exists

The account problem, solved.

01

No global card needed

Getting an OpenAI or Anthropic account from Nepal is a wall. One key here gets you past it.

02

One bill, live models

Stop juggling provider dashboards. A single plan, a single key — and a model list that only shows what works.

03

Honest limits

The model list tells you exactly what is online. Clear errors at limits, never silent charges.

04

A real person

Local support when you need it. Ask before you pay; we confirm seats first.

// from zero to first response

From zero to first response, guided.

  1. 1

    Get a key

    Start a free trial instantly — no card, no wait. Or ask in the WhatsApp group for a Pro seat.

  2. 2

    Point your SDK

    Change the base URL and key in any OpenAI SDK. Copy-paste examples in the docs.

  3. 3

    List models & ship

    Call GET /v1/models — it only shows what is actually online — then build. Your account page walks you through it.

// value check

What individual access costs vs. one gateway.

To reach frontier models directly from Nepal, you need foreign accounts, multiple payment methods, and global billing — if the providers accept your region at all.

Route Accounts needed Payment Est. monthly cost
Direct providers (US cards only) 5–7 Global credit card $50–200+
Nepal AI Gateway 1 eSewa NPR 599 / month

Nepal AI Gateway is not a discount service — it is an access layer. You pay for convenience, local payment, and unified billing, not cheaper tokens.

// developer access

Full developer access for 30 days.

50M tokens, 60 RPM, eSewa payment — and a live model list that only ever shows what your key can actually call.

Ask for access Usage guide

Payment is confirmed via eSewa. Contact support for instant activation.

NPR599/ month (30 days)
Monthly budget
50M tokens
Request speed
up to 60 / min
Context & Output
200K window
Models
Live catalog (GET /v1/models)
Rollout
25 seats total

Multi-provider failover across Nvidia NIM, OpenRouter, Cerebras, and more.

// why you can trust this

No lock-in. No hidden fees. No surprises.

We are not a black box. You get clear limits, direct support, and full transparency on what works and what does not.

Your key, your data

We do not store conversation content. We only route requests and enforce quota.

Clear usage dashboard

See exactly how much quota you have used, how much remains, and when your plan expires.

WhatsApp support

Real humans answer questions before you pay. No ticket queues. No auto-responders.

OpenAI-compatible

If you ever leave, your code still works with any OpenAI-compatible provider. Zero migration cost.

// already have a key?

Know exactly where you stand.

Sign in with your key to see models, quota, rate limit and expiry — without revealing the full key again.

Check my account

// good to know

Straight answers.

Who is this for?

Students, developers and small teams who want several frontier models through one familiar API.

Do I change my OpenAI code?

Usually just the base URL, key and model name. Docs include cURL, Node.js and Python.

Which models will my key include?

Your exact list appears in your account after activation. The account page and GET /v1/models are authoritative.

What happens at a limit?

The API returns a clear error instead of quietly charging more. Your account shows remaining quota and expiry.

What happens after 30 days?

The key stops working when the 30-day window ends. Contact support for an instant renewal.

Start building with frontier AI today.

One key. The models that are actually online. No global cards. No friction. Built for Nepal.

Join WhatsApp & ask for a seat Read the docs first

Seats are limited. We confirm availability before you pay.