Laya

A small Laya model chooses which skills, rules, agents and MCP tools each request actually needs, so Claude Code loads only those.

A Rubik's-cube agent that combines 3D state, temporal memory and calibrated decision probabilities from Laya and Jev to choose each turn.

A privacy-first command-line tool that checks Persian, English and mixed text for safety with Laya-MLX plus deterministic PII and secret detection.

A hotkey voice assistant for Linux using faster-whisper, a Laya typed-decision router, Ollama and Piper, all offline.

A voice assistant that drives a Mac with Laya, gliding the mouse, rippling clicks and offering 12 window-tiling modes with a frosted HUD.

Compares Laya MLX's typed move choices against Dijkstra on seeded weighted mazes, with a side-by-side replay on Apple silicon.

A real-time traffic simulation where on-device Laya and cloud Jev control identical cities so you can compare their decisions.

Runs the open 421M Laya model as the controller in a real-time Doom loop, a response to TypeSafe's Jev Doom demo.

Prunes Laya's vocabulary and quantizes it to int8 so it fits on small boards, evaluated on human-labelled robot commands.

A Rich-based terminal Tetris built so a local Laya MLX model can play it on Apple silicon.

A project where Laya-MLX plays the three-player Chinese card game Dou Dizhu, making each play from candidate options.

Combines screenshot perception, Laya decisions on Core ML and Qwen planning to play, click loot and review its own evidence, entirely locally.

Synthetic Chinese workplace-chat scenarios with frozen inputs, prompts and labels. Reports classification quality, mis-generated tasks and latency: Jev 253 ms vs Laya 151 ms.

Hermes plugin that classifies each task against a ~300-skill index with Laya, injects the top skill's SKILL.md as per-turn context, and fails open. Cache-safe by design.

Daniel Tremer's autonomous agent for Hordes.io: Qwen3.5-4B on MLX plans, Laya on Core ML decides, Apple Vision reads the screen. All on a 48 GB M3 Max, no cloud.

Pythongiant's OpenCode plugin continuously scores agentic drift with Laya plus Jensen-Shannon divergence, and plots drift over time. Code in pythongiant/laya-drift.

A Java worker that runs Laya's choice, score and yes/no decisions inside Camunda 8 workflows, with versioning and safety checks.

Trains the 421M Laya ModernBERT model to choose Balatro actions from candidate lists, using simulator self-play and a real-game test set.

A pixel space shooter whose ship is controlled by Laya through a local /v1/systemone API, forked from a Jev version of the same game.

Uses Laya's typed decisions to work through murdle.com's daily murder-mystery logic puzzles.

A head-only fine-tune of Laya that screens commands before execution and routes them to the right tool.

Native Windows demo with live probabilities, executed moves, inference timing and planner-intervention counts. A deterministic search proposes routes, Laya classifies them.

Backend-independent Unity client for local Laya servers and TypeSafe-compatible endpoints. Game state in, Choice/Score/Noul out; transport-first, no embedded weights.

A post pointing to a third-party Jev-vs-Laya study as evidence that task-specific fine-tuning closes, and often flips, the accuracy gap between the open Laya model and the closed Jev.