News

Latest from Pocai

Product updates, beta programs, and announcements.

Pocai Moderate 1.0
New

Pocai Moderate 1.0 — Beta Program

Our next-generation content moderation model is now in testing. Built on XLM-RoBERTa-large with 560M parameters, trained on 24,500 samples across 7 categories. 83% accuracy with 82.5% Macro F1.

83.0%
Accuracy
82.5%
Macro F1
7
Categories
7
Languages

Join the Beta — Earn Rewards

Looking for testers to improve Moderate 1.0. Contributors classifying 10+ samples correctly receive free API credits.

Community
Discuss with other testers
Classify
Categorize test messages
Earn
10 correct = free credits
Join Beta Program

More updates coming soon.