kyutai / feature-extraction updated 1 year ago

mimi

Mimi codec is a state-of-the-art audio neural codec, developped by Kyutai, that combines semantic and acoustic information into audio tokens running at 12.5Hz and a bitrate of 1.1kbps.

Params
100 M
Context
8,000
Downloads 30d
458 K
Likes
326
Commercial use: allowed cc-by-4.0 Not gated SAFETENSORS View on Hugging Face ↗

Download history

daily snapshots · 55 days
▲ 103 K in the last 30 days (18.3%)
858 K458 K
Jul 28Aug 15Sep 2Sep 20

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

FileQuantSizeEst. VRAMVerdict on RTX 4090 · 24 GB
model.safetensors f32 0.4 GB 0.9 GB ✅ Runs comfortably
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Specifications

Architecture
MimiModel
Parameters
100 M
Tensor type
F32
Context length
8,000
Layers / heads
8 / 8
Licence
cc-by-4.0
First seen on the Hub
2024-09-10
Training datasets
undisclosed
Added to our catalog
2026-07-28
Compare with any feature-extraction model