Skip to content
AI Model Radar

DiarizationLM Gemma 4 E4B v1

google/DiarizationLM-Gemma-4-E4B-v1
Open Source · chat · open-weights
Open Alert me on changes
Context
131.1K
Max output
—
Weights
Open
API $/1M
—
Modalities
text · image
Released
04 Oct 2026
License: apache-2.0 · google/DiarizationLM-Gemma-4-E4B-v1
Download image Share on X Share on LinkedIn
AI summary
● machine-written

Google releases DiarizationLM Gemma 4 E4B v1, speaker diarization model

DiarizationLM Gemma 4 E4B v1 is Google's open-weight speaker diarization model built on the Gemma 4 E4B foundation (4B dense parameters) and fine-tuned with Locality-Preserving Oracle Supervision across canonical speaker diarization benchmarks. The model supports text and image inputs under Apache 2.0 license.

What's new
  • Fine-tuned on all 4 canonical speaker diarization benchmark corpora
  • Built on Gemma 4 E4B foundation with 4B dense parameters
  • Open-weights architecture available via Hugging Face
  • Apache 2.0 licensed for commercial use
Best for
Speaker diarization tasksAudio transcription and speaker identificationOpen-source deployment on local hardware
Sources

Source: https://huggingface.co/google/DiarizationLM-Gemma-4-E4B-v1