Musubi Releases PolicyLM-1.7B for Real-Time Moderation
Musubi's open-weight PolicyLM-1.7B applies plain-English content policies in under 50 milliseconds without retraining.
Musubi announced and released PolicyLM-1.7B, a lightweight decision model for real-time content moderation, on October 6, 2026. The model was published with open weights.
According to Musubi, PolicyLM-1.7B applies a content policy written in plain English to messages in under 50 milliseconds and returns a binary judgement on whether content falls into a category or not. The company says the model is designed to be similar in cost and speed to the AI classifier systems that power moderation on most social platforms.
Musubi says the model can apply complex policies without special training and does not require new training when a policy changes. Co-founder and chief AI officer Filip Jankovic said PolicyLM-1.7B gives platform managers a way to label content proactively.
Earlier decision models
The release follows Typesafe AI's decision model Jev, which arrived in September 2026. OpenAI and Amazon released competing decision models shortly after Jev's release. Jankovic's interest in decision models traces back to a 2024 project called GLiNER.
Quick answers
What is PolicyLM-1.7B?
It is a lightweight decision model for real-time content moderation, released by Musubi with open weights. It applies a content policy written in plain English and returns a binary judgement.
How fast is PolicyLM-1.7B?
Musubi says it applies a content policy to messages in under 50 milliseconds, with cost and speed similar to the AI classifiers used on most social platforms.
Does PolicyLM-1.7B need retraining when a policy changes?
No. Musubi says the model can apply complex policies without special training and does not need new training when the policy changes.