TypeSafe AI's Jev is a decision model designed for agentic security workflows, answering fixed-choice questions with probabilities instead of generated text. In a proof of concept for alert triage, Jev filtered 15-33% of alerts with 98% accuracy while reducing latency and cost by handling multiple decisions in parallel at 70x cheaper per token than frontier LLMs.
LingBot-World 2.0, a 1.3B world model, achieves 16.1 FPS on a single RTX 5090 through inference optimizations, delivering 2.7× speedup over the original paper's implementation while maintaining lossless performance. The optimizations involve fused kernels and precision techniques applied to the upstream model without altering its architecture.