GitHub radar
Get a probability for each answer instead of text — fast and free
This model skips writing text and returns a calibrated probability for every answer option in under half a second. You get a ready decision, not a paragraph someone has to parse by hand.
It's a decision-making AI model, not a chatbot: it never writes text, it just returns a probability for each answer option to your question.
When it helps
Useful when you need to quickly sort emails, tickets, or comments into ready-made categories — department, urgency, churn risk. For questions with dozens of options, pick another tool: it confuses similar choices.
Pros
- Answers in under half a second even on a plain CPU.
- Works across 100+ languages out of the box.
- Apache 2.0 license — self-host it for free.
- About 7.8x faster than the paid Jev service per question.
Cons
- Confuses similar options once there are 20 or more.
- Fine-tuning is almost required for serious accuracy.
How to set it up — step by step
- 1Open your AI agent (Claude Code, Codex) inside your project folder.
- 2Send the agent the text from the block below.
- 3Give the agent a few test support messages to classify.
- 4Check the probabilities for each question against what you expected.
- 5If accuracy falls short, ask the agent to fine-tune it on your own data.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up Convai Innovations' Laya model for me from https://huggingface.co/convaiinnovations/laya — run it locally, feed it a few sample support tickets (department, urgency, churn risk), and show me the probability it returns for each question.
A regular laptop with 8 GB of RAM or more — no GPU needed, answers come back in well under half a second on CPU alone.
Open on Hugging Face▌ More finds