Laya on MLX
laya-mlx is an independent MLX port of Laya that runs the original weights on the Apple Silicon GPU with no PyTorch or Transformers runtime. It keeps upstream prompt formatting, calibration and output schema, and it is the base for many Mac demos, agents and benchmarks.
260 projects
Bud-Decision-Engine
BudEcosystem
Desktop app and local server for Laya and other Jev-like decision models
laya-mcp
wsargent
Runs local Laya inference on Apple Silicon and exposes typed-decision tools through an MCP server
laya-router-skill
wangmiaozero
Adds local Laya decision routing to coding agents through skills and MCP
doomLaya
azalio
Trains and compares Laya and Jev agents playing FreeDoom with reproducible results
oh-my-laya
leo1394
Installs local Laya tools for classification, scoring, and routing in coding agents
laya-MLX-8bit
cavi-ai
An 8-bit MLX conversion of Laya checkpoints for Apple Silicon
localjev-mlx
rimusz
A self-hosted System One judgment service for Apple Silicon using laya-mlx
doom-war
rythmn1111
Compares local MLX Laya and hosted Jev as decision-makers in a Doom deathmatch
laya-vs-jev
zaferayan
Benchmarks multilingual Laya against hosted Jev across 900 cases, three tasks, and six languages
SnakeGame_Laya
Okbatti
Uses a fine-tuned Laya model to choose moves in a Snake game
laya-mlx-zh
ZLHAOOO
Provides Chinese fine-tuned Laya weights, training data, and evaluations for Apple Silicon MLX
design-os-generative-ui
jangtrinh
Builds a generative UI engine using local Laya-MLX decisions and a TypeSafe JEV cascade router
pacman-arena
hakan-gecili
Compares Laya with other local decision models playing Pac-Man under identical conditions
s1-tui
tnaftali
Provides a terminal interface for typed decisions using local Laya or the hosted Jev API
laya-modernbert-decision-90pct
ameerhmz5
Fine-tunes Laya into a ModernBERT decision model and offers deployment formats
Laya-Multilingual-MXFP8
sahilchachra
An MXFP8 MLX quantization of Laya multilingual's encoder backbone for Apple Silicon
laya-toolguard-mlx
a3harva
Fine-tunes Laya in MLX to assess coding-agent tool calls for safety risks
laya-multilingual-arc
goktugoguz
Fine-tunes Laya to classify Arc USDC transfers and answer questions about them
fin-laya-mlx-6-rounds
ya9919
A 0.42B-parameter financial decision model trained on CHFS, CGSS, CFPS and CLDS data
laya-multilingual-stellar
goktugoguz
Fine-tunes Laya to classify Stellar operations and answer questions about them
laya-mlx
millat
Converts Laya to a native MLX FP16 checkpoint for inference on Apple silicon
LayaVoxprompt
MainzelMennchen
Fine-tunes multilingual Laya to classify dictation as an assistant prompt or ordinary text
laya-snake-mlx
madhavbiplov
Fine-tunes a multilingual Laya model for calibrated snake-game decisions on Apple silicon
fin-laya-mlx-3-rounds
ya9919
A 0.42B-parameter financial decision model trained on CHFS and CGSS data
Install
Requirements from the laya-mlx README: Apple Silicon, Python 3.11+, macOS 14+.
pip install laya-mlx
Pre-converted FP16 checkpoints are on Hugging Face: aac6fef/laya-mlx, aac6fef/laya-multilingual-mlx and aac6fef/laya-typed-decisions-mlx. You can also load the original convaiinnovations/laya IDs.
Router and CLI
The port includes the upstream language router and presets:
from laya_mlx import Router, triage_questions
router = Router(dtype="float16", max_loaded=2)
result = router.predict({"message": "发票被重复扣款，请退款。"}, triage_questions())
print(result["routing"]) # multilingual
For the terminal demo, where Laya plays Snake:
pip install 'laya-mlx[demo]'
hf download aac6fef/laya-multilingual-mlx
laya-snake
Reported numbers
On an M3 Max, FP16, one short question: 13.42 ms P50 for the 421M English model and 7.39 ms for the 322M multilingual model. The README states that all three checkpoints matched the upstream answer on 63/63 validation questions in FP32 and FP16. That measures port fidelity, not task accuracy.
Projects built on laya-mlx
- laya-ultrafast: a port of browser-use/jev-ultrafast that makes browser decisions locally.
- laya-mcp (wsargent): an MCP server with triage, guard, moderation and email tools.
- laya-fast: adds Neural Engine execution next to the MLX GPU.
- laya-jev-lab and jev-laya-benchmark: independent Jev comparisons that use laya-mlx as the Laya side.
- laya-vs-jev: Laya and Jev playing Chrome's T-Rex game side by side.
Caveats
This is not an official Convai Innovations release. Training and fine-tuning stay in the upstream project. BF16 can be requested but is not part of the published validation.
More ways to use Laya