# OneToken API

> OpenAI- and Anthropic-compatible AI API gateway with ruble billing via SBP (Faster Payments System) and direct access without VPN.

OneToken provides developers with a single unified API key (`sk-...`) to access frontier AI models from Anthropic, OpenAI, DeepSeek, and Google. It is engineered for low latency, zero monthly subscription fees, pay-as-you-go per-token billing, and seamless compatibility with coding agents (Cursor IDE, Cline, Roo Code, Claude Code) and standard client libraries (`openai`, LangChain, LlamaIndex).

- **Website**: https://onetoken.ru
- **OpenAI-Compatible Base URL**: `https://api.onetoken.ru/v1`
- **Anthropic-Compatible Base URL**: `https://api.onetoken.ru`
- **Management & Top-up**: Telegram Bot [@OneTokenRobot](https://t.me/OneTokenRobot)
- **Status / Health check**: `https://api.onetoken.ru/api/status`

---

## Key Features

1. **Direct Access without VPN**:
   Clients connect directly to `api.onetoken.ru`. The gateway routes requests through a high-bandwidth European network contour with low latency to AI provider backends.
2. **Pay-as-you-go (No Subscriptions)**:
   No recurring $20/month fees that expire unused. Users pay only for actual input and output tokens consumed. Balance never expires.
3. **Instant Ruble Top-up via SBP**:
   Deposits start from 100 ₽ via QR-code or banking app link, plus Russian bank cards.
4. **Welcome Test Balance**:
   New users receive a 15 ₽ starter grant immediately upon issuing a key in [@OneTokenRobot](https://t.me/OneTokenRobot) (enough for ~400,000 DeepSeek tokens).
5. **Full Protocol Support**:
   Unbuffered streaming (Server-Sent Events / SSE), Tool Use / Function Calling, Structured JSON Outputs, Multimodal Vision, and extended reasoning timeouts (up to 300s for DeepSeek R1).

---

## Available Models & Pricing (RUB per 1,000,000 Tokens)

| Model ID | Provider | Input Price (1M tok) | Output Price (1M tok) | Best For |
| :--- | :--- | :--- | :--- | :--- |
| `claude-sonnet-5` | Anthropic | 170 ₽ | 850 ₽ | Industry benchmark for architecture and coding agents |
| `gpt-6-astra` | OpenAI | 160 ₽ | 800 ₽ | Next-gen reasoning, coding, and multimodal tasks |
| `deepseek-v4-pro` | DeepSeek | 33 ₽ | 131 ₽ | Olympiad-level code, math, and deep reasoning |
| `grok-4.7` | xAI | 100 ₽ | 300 ₽ | Real-time web knowledge, fast processing |
| `gemini-2.5-flash` | Google | 12 ₽ | 45 ₽ | Ultra-fast responses (300 ms), 1M context |
| `claude-opus-5` | Anthropic | 300 ₽ | 1 500 ₽ | Maximum intelligence for complex architectural audits |

*Full and current model catalog can be fetched via `GET https://api.onetoken.ru/v1/models` using your API key. Classic aliases (`claude-3-5-sonnet`, `gpt-4o`, `deepseek-chat`) are routed automatically.*

---

## Quickstart & Integrations

### 1. Cursor IDE
In Cursor settings (`Ctrl/Cmd + Shift + J` -> **Models**):
- **OpenAI API Key**: `sk-ot-...` (obtained from @OneTokenRobot)
- **Override OpenAI Base URL**: `https://api.onetoken.ru/v1`
- Click **Verify**, then add models: `claude-sonnet-5`, `gpt-6-astra`, `deepseek-v4-pro`.

### 2. Cline / Roo Code (VS Code Extension)
In Cline Settings:
- **API Provider**: `OpenAI Compatible`
- **Base URL**: `https://api.onetoken.ru/v1`
- **API Key**: `sk-ot-...`
- **Model ID**: `claude-3-5-sonnet` or `deepseek-chat`

### 3. Claude Code (CLI)
Set environment variables:
```bash
export ANTHROPIC_BASE_URL="https://api.onetoken.ru"
export ANTHROPIC_API_KEY="sk-ot-..."
claude
```
On Windows PowerShell:
```powershell
$env:ANTHROPIC_BASE_URL="https://api.onetoken.ru"
$env:ANTHROPIC_API_KEY="sk-ot-..."
claude
```

### 4. Python (`openai` SDK)
```python
from openai import OpenAI

client = OpenAI(
    base_url="https://api.onetoken.ru/v1",
    api_key="sk-ot-...",
)

stream = client.chat.completions.create(
    model="deepseek-chat",
    messages=[{"role": "user", "content": "Write a debounce helper in TypeScript."}],
    stream=True,
)

for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="", flush=True)
```

### 5. Node.js (`openai` package)
```javascript
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.onetoken.ru/v1",
  apiKey: "sk-ot-...",
});

const response = await client.chat.completions.create({
  model: "claude-3-5-sonnet",
  messages: [{ role: "user", content: "Explain async iterators in JS." }],
});

console.log(response.choices[0].message.content);
```

### 6. Raw cURL
```bash
curl https://api.onetoken.ru/v1/chat/completions \
  -H "Authorization: Bearer sk-ot-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-chat",
    "messages": [{"role": "user", "content": "Ping"}]
  }'
```

---

## Authentication & Security

- API keys follow the format `sk-ot-...` and can be regenerated at any time inside [@OneTokenRobot](https://t.me/OneTokenRobot).
- User data is not used for model training. All traffic is encrypted over TLS 1.3.
- Only billing metadata (model, tokens, timestamp) is logged for balance accounting.

## Support & Contacts

- **Telegram Bot**: [@OneTokenRobot](https://t.me/OneTokenRobot)
- **Community Channel**: [@OneTokenVibe](https://t.me/OneTokenVibe)
- **Email Support**: support@onetoken.ru
- **Gateway Endpoint**: `api.onetoken.ru`