Skip to content

granite-4.0-1b-speech-8bit

mlx-community/granite-4.0-1b-speech-8bit
Uncategorised · mlx-audio
8-bit Open on Hugging Face
Parameters
2.3B
Weights (est.)
≈ 2.9 GB
Mac memory (est.)
8 GB+

Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.

Details

Base model
ibm-granite/granite-4.0-1b-base (quantized)
Vendor
IBM · Granite
Task
Uncategorised
Quantisation
8-bit
Licence
apache-2.0
Downloads
133 in the last 30 days · 1.1K all time · 0 likes
Published
2026-03-11 · updated 2026-03-11 (8mo ago)
Languages
multilingual, en, fr, de, es, pt, ja
Tags
mlx-audio, granite_speech, speech-to-text, speech, transcription

Other MLX builds of ibm-granite/granite-4.0-1b-base

Build Quantisation Weights Mac memory Downloads
granite-4.0-1b-base-3bit 3-bit ≈ 714 MB 8 GB+ 21
granite-4.0-1b-base-4bit 4-bit ≈ 918 MB 8 GB+ 41
granite-4.0-1b-base-5bit 5-bit ≈ 1.1 GB 8 GB+ 29
granite-4.0-1b-base-6bit 6-bit ≈ 1.3 GB 8 GB+ 23
granite-4.0-1b-base-8bit 8-bit ≈ 1.7 GB 8 GB+ 30
granite-4.0-1b-speech-4bit 4-bit ≈ 2.0 GB 8 GB+ 70
granite-4.0-1b-speech-mxfp4 MXFP4 ≈ 2.0 GB 8 GB+ 18
granite-4.0-1b-speech-5bit 5-bit ≈ 2.2 GB 8 GB+ 51
granite-4.0-1b-speech-6bit 6-bit ≈ 2.4 GB 8 GB+ 24
granite-4.0-1b-speech-8bit (this one) 8-bit ≈ 2.9 GB 8 GB+ 133
granite-4.0-1b-base-bf16 bf16 ≈ 3.3 GB 8 GB+ 22
granite-4.0-1b-speech-bf16 bf16 ≈ 4.6 GB 8 GB+ 39