llama.cpp Adds Support for Nemotron 3.5 Transcribe

·2026.11.03 23:13

Key point

The model features a hybrid mixture-of-experts architecture activating approximately 4.3B parameters and supports 40 languages.

llama.cpp pull request #31471 adds support for Nemotron 3.5 Transcribe.
Nemotron 3.5 Transcribe is an open-weight, offline speech and audio model for transcription in 40 languages.
The model supports recordings up to one hour with a 65,536-token context.
It uses a hybrid mixture-of-experts architecture activating approximately 4.3B parameters, including the audio encoder.
Released in BF16, FP8, and NVFP4 precision.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.