Typed clients that retry, wait and verify for you.
OpenAI-compatible API gateway for Claude, GPT, Gemini and 50+ models without VPN from Russia
Stage
Not given
Startup
Founded
Not given
2025
Based in
Not given
SG
Pricing model
Not given
Paid
Pricing
Not given
Pay-as-you-go per-token pricing with no subscription or minimum spend. Minimum top-up $5 in USDT/USDC or foreign Visa/Mastercard. Prices vary by model: DeepSeek Flash from $0.154 per million input tokens, Claude Sonnet 5.5 at $2.20 per million input tokens, GPT-5.4 at $2.75, Gemini 2.5 Flash at $0.3
Key features
Automatic retry with exponential backoff for 429 and 5xx errors
Idempotency key handling to prevent duplicate prints on retry
Job polling with jobs.wait() method
Webhook signature verification
Support for Node 18+, Deno, Bun and edge runtimes
Synchronous and asynchronous Python client
No external dependencies for JavaScript
Python SDK uses only standard library
TypeScript support with full type definitions
Support for PDF/A, PDF/X and PDF/UA compliance
OpenAI-compatible API endpoint for unified access
Native Anthropic Messages API support for Claude Code
50+ models from 10 providers including OpenAI, Anthropic, Google, DeepSeek, Qwen
Prompt caching support reducing costs 10x for cached context
Direct access from Russia without VPN requirement
Payment in USDT/USDC or foreign-issued Visa/Mastercard cards
Models at native precision without additional quantization
No subscription required, pay only for tokens used
Supports streaming responses and tool calling
Data not used for training unlike some aggregators
What makes it different
Automatic retry handling with Idempotency-Key to ensure requests are processed exactly once
No external dependencies, using only platform fetch and Web Crypto for JavaScript
Complete typing support in TypeScript and documentation in Python
Unified error type with detailed status, code, title, detail and request ID
Access Claude and other restricted models directly from Russia without VPN or Anthropic account
Single OpenAI-compatible endpoint for models from 10 different providers
Pay-as-you-go pricing without subscription or minimum spend requirements
Genuine models at native precision without quality reduction or substitution
Integrated prompt caching for 10x cost reduction on repeated context