Laya

by Convai Innovations

A small free decision model (Apache 2.0): it answers with a choice, a score or a yes/no with probability, in tens of milliseconds on your own PC.

Apache-2.0 FreeOpen sourceOpen weightsLocal

View on GitHub →

What it does

322 to 421 million parameters, 100+ languages, no per-call cost
Same kinds of questions as Jev (choice, score, noul) in a single pass
Fine-tune it on your data: out of the box, accuracy is well below Jev

! Free alternative to Jev (Apache 2.0), released September 18, 2026. Note: its base models are near chance without tuning; you need to fine-tune it on your own examples (the best published score comes from a version already tuned on the benchmark). On an independent 49-task benchmark: 0.58 accuracy against 0.97 for Jev. Information from the official repository and tech press.

Install it

Requirements: Python and pip. A CPU is enough; weights download from Hugging Face on first run.

Install the package

pip install laya

Load the model

python -c "import laya; agent = laya.load('convaiinnovations/laya')"

The prompt for your agent

Paste it into your coding agent: it checks the requirements and installs it for you.

Claude CodeCodexCursorGemini CLIopencode

Prompt

Install Laya (https://github.com/NandhaKishorM/laya) on my machine.
1. Check the requirements: Python and pip. A CPU is enough; weights download from Hugging Face on first run. If anything is missing, tell me before installing anything.
2. Follow the official method:
- Install the package: pip install laya
- Load the model: python -c "import laya; agent = laya.load('convaiinnovations/laya')"
3. Check that it works, then explain how to launch it and where to start.

Always review what your agent is about to run, especially install commands.