Laya

Tests whether the open Laya model on Apple's MLX can stand in for a paid judge, scoring model outputs locally with typed answers.

Two LLM routers behind one endpoint compared on 180 labelled requests: Laya picks small, medium or powerful in one forward pass versus a GPT-5 nano reasoning router. Local and free for the Laya side.

Independent Jev-vs-Laya measurements plus a local-first cascade: Laya decides first, high-confidence answers execute, low-confidence ones escalate to Jev. Reported as matching Jev's accuracy at lower cost.

Tobias built his own Snake benchmark and ran the Jev API against Laya-MLX and Kev-4B on a low-RAM MacBook. Honest take: Jev wins on quality, locals win on speed.

Yankis's video runs Nimble, Decider, OpenJev, Laya and more through the same JevBench scoring. Laya gets the final 3:41.

Speak commands to drive a browser on Apple silicon, with a local Laya decision model checking each result before the next action.

A studio for fine-tuning Laya typed-decision models locally, with checks that prove the tuned model beats the baseline before you ship it.

Allen Porter's native Assist conversation agent scores intents and entities in one Laya pass, runs device commands locally and escalates open-ended queries to a fallback LLM.

Shantanu Goel trained Laya for ~24 hours to play FreeDoom E1M1 from text-only perception. Jev played well zero-shot; the Laya base model did not. Honest negative result.

Six classifiers over SST-2, AG News and more, reporting accuracy, macro-F1, ECE, p50/p95 latency, throughput and cost side by side. Laya runs locally on MPS or CPU.

AmirTaha wrote a Snake game that feeds fruit, head and border positions to a local Laya model and takes the next direction from it. Persian post, code in AmirTahaMim/LayaSnakeGame.

A browser agent that lets Laya choose among indexed on-page actions locally, completing a Google Flights search demo in about 6.5 seconds.

An MCP server and skill that gives the oh-my-pi coding agent typed decisions from a local Laya model in milliseconds and with zero LLM tokens.

Paper-trades Binance crypto and S&P 500 stocks with Laya MLX, with a live dashboard, backtests, a prompt lab and walk-forward tests, all on a Mac.

Hope put both models in a 1v1 Doom deathmatch, first to five kills. Jev's calls were sharper and needed a third of the corrections; Laya decided twice as often.

Chinese fine-tuned Laya weights for MLX that score message routing, urgency, priority and retrieval keywords in one forward pass of about 27 ms.

Unmodified Laya weights, real NES emulation via Stable Retro, text observations from RAM, model-driven jumping. 27.5 s of game time at 16.79 ms median inference.

Dockerized FastAPI service bundling all three checkpoints behind a language router, with timing-safe API keys, OpenAPI docs, triage and moderation presets and bulk inference.

Anton Lobach ported the Laya System 1 engine to Jolt, a Clojure implementation on Chez Scheme. Code in jlt-commons/lev.

A Pi coding-agent extension that uses a local Laya or remote Jev decision model to judge task difficulty and switch to a cheaper or stronger LLM.

Combines tree-sitter, BM25 and a Laya re-ranker, delivered through hooks and MCP, so Claude Code pulls the most relevant code first.

One process serves the model API and an admin UI; agents call decide over MCP, HTTP or CLI and take top.id as the action. Windows service installer included.

An experimental mod that lets the open-weight Laya model manage a RimWorld colony autonomously on the player's own machine.

One typed question per frame; the other paddle is three lines of arithmetic as a control. Physics in WebAssembly, decisions in native Laya over POST /decide.