AOS GLM is a deterministic language engine written in C++20 with SIMD acceleration, offering a drop-in replacement for TypeSafe Jev and Instructor APIs. It operates entirely in host memory with sub-microsecond latencies and includes Python bindings, REST API, and CLI interfaces for binary decisions, taxonomy classification, and severity scoring.
Oryn is a compact, low-latency decision model based on BERT Mini that performs classification by scoring dynamically supplied options at inference time rather than using a fixed output layer. At approximately 21 MB with 3 ms latency, it handles binary, multiclass, and ordinal tasks and was trained on phishing detection, reading comprehension, news classification, and sentiment analysis datasets. The model is optimized for local, CPU-friendly inference and works best on tasks similar to its training data.
TypeSafe released Blink, a 452KB decision model in C/WebAssembly that answers structured questions in a single forward pass with no token generation or output decoding. The embeddable runtime has zero allocations during scoring, runs on minimal memory, and builds to a 66KB WebAssembly module for browsers and Node without external dependencies.
Jev is TypeSafe AI's decision-making model that answers typed questions in parallel within milliseconds, returning calibrated probabilities instead of generating text. JevMade is a community registry showcasing 743 verified experiments built with Jev across games, agents, routing, and automation tasks.