Open Source
Built in the open
We're passionate about machine learning, AI and Rust, and we love what happens when they meet. Here are the inference engines, tools and frameworks we build along the way, free for everyone to use, study and improve.
h3-rs
A Rust video generator: turn a text prompt into video, running natively.
drama-rs
A Rust inference engine for quantized Drama-3 (SOTA voice synthesis).
fluid-speech-rs
A Rust port of Fluid Speech Mini inference for low resources SOTA voice transcription.
julia-rs
A Rust port of Julia-1 inference runtime.
laya-rs
Rust reimplementation of Laya, a sub-35ms non-autoregressive typed-decision engine, built from scratch on candle.
gliner-rs
Rust inference for GLiNER2 checkpoints. Entity/relation extraction, classification, and structured JSON extraction.
imajev-rs
A Rust image classifier: fast, lightweight inference for labelling images.
opendlss-rs
A Rust port of DLSS5 inference (Vulkan + CUDA).
locate-anything-rs
A Rust port of LocateAnything inference.
sam-rs
Rust inference for the Segment Anything family: SAM, MobileSAM, EfficientSAM, SAM 2 and SAM 3.
idol-rs
Inference engine for Idol: turn a single image into a 3D Gaussian Splat model.
merge-safetensors-rs
A Rust CLI to merge safetensors.
nvidia-smi-live
A real-time, live-updating CLI & web dashboard for monitoring NVIDIA GPU metrics.
portward
A tiny local web server that shows which app is listening on which port on your machine.
apiplant
Ship one executable, point it at a directory, get a database-backed REST API.
- and more here!