OpenAI-compatible AI API relay for multiple AI models with prepaid wallet billing
Open-source engine for faster local language model inference with bit-for-bit identical output
Stage
Not given
Not given
Founded
Not given
Not given
Based in
HK
Not given
Pricing model
Paid
Open source
Pricing
Chat models from $0.019708 per million tokens; Image models from $0.00386 per image; Video models from $0.016 per second; Audio models from $0.00291 per request; Embedding models from $0.019 per million tokens. Example rates: Grok 4.6 at $1.9 input/$5.7 output per million tokens, Claude Opus 5.5 at
Not given
Key features
OpenAI-compatible base URL and SDK support
197 AI models across 30 brands
Chat, image, video, audio and embedding models
Prepaid wallet billing with per-token pricing
Minimum $1 top-up with multiple payment methods
Alipay, Stripe, Antom and cryptocurrency payments
Dashboard with usage and spending records
Seven-language interface
API keys and spending limits management
Public API status page for provider availability
Speculative decoding for faster token generation
Prompt caching for context reuse
Real-time prompt processing while typing
Concurrent request batching
Bit-for-bit identical output verification
Multi-GPU and multi-backend support (CUDA, Vulkan)
Vision model support with image encoding
Mixture-of-experts model support
RAD-1 automatic draft token generation
Cross-platform support for Windows and Linux
What makes it different
Single API key and endpoint for 197 models from 30 brands
OpenAI SDK compatibility with one-line URL change
Global payment options including cryptocurrency and Alipay
Transparent per-model pricing visible before API calls
Support for emerging AI providers and latest models
Exact deterministic inference: all default optimizations preserve bit-for-bit identical output compared to plain decoding
4.5× faster performance on editing tasks on small models compared to llama.cpp
Verifiable reproducibility through receipt system that records all request parameters
Nine optimization techniques enabled by default with ability to disable each individually
No dependency on draft models or external selections for speculative decoding
TokenMix
Kinetic