Model Tracker

EnCodec

EnCodec is an AI model made by Meta. It is generally available. It takes audio and returns audio. Its weights are open. Its licence is MIT.

Released: Not confirmed: two readings disagree: 2022-10-24 and 2022-10-25
Status: Generally available — github.com, read 16 October 2026
Takes: Audio — github.com, read 16 October 2026
Returns: Audio — github.com, read 16 October 2026
Open weights: Yes — github.com, read 16 October 2026
Licence: MIT — huggingface.co, read 16 October 2026
Context window: Not confirmed: The maker's pages describe an audio codec and do not print a context window in tokens.
Longest answer: Not confirmed: The maker's pages do not state a standard maximum output length in tokens for EnCodec.
Knowledge cutoff: Not confirmed: The maker's pages do not state a knowledge-cutoff date for EnCodec.
API price, input: Not confirmed: The maker's pages do not publish a standard API input-token price for EnCodec.
API price, output: Not confirmed: The maker's pages do not publish a standard API output-token price for EnCodec.
Where: Downloadable weights — github.com, read 16 October 2026
Parameters: Not confirmed: The maker's model cards list different parameter counts for the two checkpoints (23.3M for 24 kHz and 19.1M for 48 kHz); the register field allows only one number for EnCodec.
Model card: https://huggingface.co/facebook/encodec_24khz — huggingface.co, read 16 October 2026

Questions

Yes: it is generally available, as Meta states it.

Yes: Meta publishes the weights of EnCodec for download.

The licence of EnCodec is MIT, as Meta names it.

EnCodec takes audio and returns audio.

AudioCraft - Meta AI (read 16 October 2026)
EnCodec: High Fidelity Neural Audio Compression (read 16 October 2026)
High Fidelity Neural Audio Compression (read 16 October 2026)
Model Card for EnCodec (read 16 October 2026)
Model Card for EnCodec (read 16 October 2026)