Route a support message
Choose a team from a short ticket and flag a deadline in the same request.
SHORT TEXT · TYPED ANSWERS
Classify a message, flag urgency, or score feedback. Try a real request below.
“Checkout fails after I enter my card. I need this order today.”
PLAYGROUND
Pick a model for your input, then an example or your own text. Which model should I use?
Short textA sentence or a short message
Long contextThreads, documents and JSON records
MediaImages, audio and short video
Attach an image, audio clip or short video and list the answers to choose from.
FOCUSED JOBS · TWO MODEL VARIANTS
Small decisions are easier to ship when the answer has a predictable shape. Try these cases in the playground above.
Choose a team from a short ticket and flag a deadline in the same request.
Classify a brief comment and flag cases for a human moderator.
Score a compact review and identify the issue it mentions.
Find the main request and whether a prospect gives a timeline.
Use Laya Multilingual for short messages such as a Chinese payment request.
THE CONTRACT
Laya reads short text with explicit typed questions. The model does not write long-form prose. Oversized requests are rejected before silent truncation can hide missing context.
NEXT STEPS
Use the source and model cards for local setup, or compare the same question across decision models.
Find the upstream Python package, open weights, and an independent Apple silicon port.
Read the local setup guide →Laya for short text, Jev or Mercury Decide for long context, Jev-Omni for images, audio and video.
See which model fits →Keep one input and question fixed, then inspect the answers on your own cases.
Read the comparison guide →COMMON QUESTIONS
Laya AI is a playground and API for short-text decisions using the Laya models. It returns typed answers rather than chat prose.
Laya English and Laya Multilingual for short text, Jev and Mercury Decide for longer text and JSON, and Jev-Omni for images, audio and video. The playground shows which models are online.
The service rejects input that exceeds the model token budget rather than silently truncating it.