A tool that analyzes Hacker News comments for sentiment, sarcasm, emotion, toxicity, and usefulness using Laya, a local System One decision model. It fetches random comments via the Hacker News API and scores them without requiring an LLM, API keys, or internet after the initial model download.
Laya is an open-weight model optimized for Apple Silicon that runs locally on macOS with Core ML, achieving 49-50 decisions per second in a Snake game demo with no generated tokens. The multilingual variant processes decisions in ~5ms on M3 Max with 2.78× better energy efficiency than MLX, and requires no PyTorch or external dependencies for inference.
A developer chose Laya, an open-weight decision model, over TypeSafe's hosted Jev service for a coding agent's small classification tasks. Laya runs locally on a Mac Studio with privacy and control benefits, while Jev offers a managed API with lower operational overhead but higher costs and less customization.
A head-to-head benchmark compares Laya and Jev language models on 751 identical test cases across 9 suites, finding Jev outperforms on multi-class and non-English tasks (intent 0.975 vs 0.725, toxic 1.000 vs 0.767) while Laya wins on agnews and mnli with zero cost and lower latency (180–660 ms vs 925–1068 ms). Emotion classification is weak on both models near 0.55 accuracy; gating at 0.85 confidence keeps 58% of Laya traffic at 0.878 accuracy and 78% of Jev at 0.917.
Laya (OS Jev) is running on Mac M4 with CoreML offline, achieving 45 decisions per second. The content references a GitHub file that can be saved to a computer and used in GitHub Desktop.
Laya MLX is an open-weight decision model running natively on Apple Silicon, delivering typed decisions in 13.4 ms median latency with zero output tokens. It enables local inference without external dependencies, demonstrated through a Snake game where every move triggers real-time decision-making with safety layer corrections.
A researcher who developed a non-autoregressive decision model in 2025 releases Laya, an open-source alternative to TypeSafe AI's Jev, which applied similar concepts without open-sourcing. Laya runs 6-8 times faster than Jev using bidirectional encoders, supports 100+ languages, and is fully open-source under Apache 2.0.