One key for production AI models.

One API key connects Claude, GPT, Gemini, DeepSeek and other models through familiar Anthropic- and OpenAI-compatible APIs.

Buy and top up in the Telegram bot or in your web cabinet

MANY MODELS. ONE WAY IN.

01 / PRICING

Pay less. Get more.

A small idea first. Something bigger next. Choose a balance for the work, not a subscription just in case.

you pay of the balance on top

Savings

Each $1 paid

of balance

Next tier at $500: each $1 buys $8.33

02 / PRODUCTION AT SCALE

A service already carrying the load.

Anthropic + OpenAI compatible Frankfurt · TLS 1.3

1. 967M+ requests served
2. $ 4.59M API balance funded
3. 88 production models · one key
4. up to 90% savings vs direct

03 / MODELS

Real use cases. Current model rates.

Every model and price comes from the current production catalog.

04 / CONNECT

Connect your tools

Ready-to-use settings for your tools. Add your key and keep working in a familiar client.

• Anthropic- and OpenAI-compatible APIs
• One public endpoint
• Ready-to-use setup guides

 export ANTHROPIC_BASE_URL= "https://api.oneprovider.dev" export ANTHROPIC_AUTH_TOKEN= "$ONEPROVIDER_KEY" export ANTHROPIC_MODEL= "claude-sonnet-5-5" claude 

Install Claude Code and set your own key in ONEPROVIDER_KEY first.

05 / FEATURES

Built for developers who get things done.

06 / ASK ABYSS

Ask Abyss. Answers that build themselves.

1. Search or Research A sourced answer in seconds, or a full report with sections and tables in a few minutes.

2. Every fact cited Numbered sources under every answer. Tap a chip to open and check the claim.

3. Charts, tables, calculators Answers come as a chart, a table, a diagram or a working calculator, not only as text.

4. Web and Telegram The same chats, memory and balance in the web cabinet and as a Telegram Mini App.

app.oneprovider.dev · @oneprovider_robot

Animated example in a phone: the OneProvider bot in Telegram greets you with its welcome video and menu, the Ask Abyss button opens a Mini App that connects to your cabinet, a question is typed and the answer builds with sources and a chart or a calculator.

07 / FAQ

Common questions.

An API lets your code, editor or agent call models like Claude or GPT directly, with no website subscription. You get an API key, connection details and a cabinet in the Telegram bot and on the web. Billing is pay-as-you-go: you top up one balance, each request is charged at the model’s per-token rate, there is no auto-charge and the balance does not expire. When it runs out, requests return a 429 “Insufficient balance” error until you top up.

OneProvider Ltd. (Cyprus) runs the service; the gateway is hosted in Frankfurt with TLS 1.3. Under our policy prompt text is not stored. Terms, refund and privacy policies are public, the status page shows live uptime, and every request is listed in your usage log. The balance is prepaid: you spend only what you top up. Legal

One key works with Claude Code, Cursor, n8n and other clients. Claude Code and Anthropic SDKs use https://api.oneprovider.dev, OpenAI-compatible tools use https://api.oneprovider.dev/v1. To switch models, change the model name in a compatible client. The documentation has setup for each client. Documentation

It can replace a gateway like OpenRouter when your tool accepts a custom base URL. You get one key for Claude, GPT, Gemini, DeepSeek and more, OpenAI-compatible and Anthropic-native endpoints (Claude Code is supported natively), prepaid packs, and a Telegram bot plus web cabinet. Compare rates on the pricing page. Migration guide

The catalog lists the current Claude models, such as Opus 5 and Sonnet 5.5, next to GPT, Gemini, DeepSeek and others, with context size and rates per million tokens, including cache rates. The rates come straight from the production catalog, so use the pricing page for current numbers. Models and pricing

YOUR IDEA WILL TAKE SHAPE

Ready to start building?

The tools are ready. Make your first move.