Deploy GPU applications on supported community hardware and cloud capacity
Open-source engine for faster local language model inference with bit-for-bit identical output
Stage
Startup
Not given
Founded
2026
Not given
Based in
SA
Not given
Pricing model
Paid
Open source
Pricing
One published price per GPU model in USD per GPU-hour. Hosts keep 90% of completed rental fees and Petabyte keeps 10%. A reservation holds funds, with final charges and returned balance following settlement after the run completes.
Not given
Key features
Choose from available GPU hardware with published hourly rates
Review cost estimates before deployment starts
Start from ready-made template applications
Monitor workload lifecycle states, logs and outputs
Connect your own GPU hardware as a host
Escrow-based payment system for secure transactions
Support for inference model requests
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
Escrow payment ensures work is delivered before host is paid
Transparent pricing: all rates published upfront with no surprises
Minimal transaction fees: hosts keep 90% of rental earnings
Automatic refunds for failed rentals on platform side
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
Kinetic