Model reference · open weights

suna

Audio jjcoders Music / audio 1 build Open weights 673 dl/mo

suna is an open-weight audio or speech model from jjcoders. suna-v1 (FP16) weighs 1.1 GB; the smallest configuration that runs it is RTX 3060 12 GB.

Request Rent a GPU

For self-hosting, compare this build's requirements below with the GPU you plan to rent.

Suna is a 2.0B parameter text-to-audio model developed by jjcoders and published as jjcoders/suna-v1.
It functions as an AI music production and vocal co-producer system that analyzes raw human singing to generate a key and tempo locked multi-instrument band with cinematic sound effects.
The model is available under the apache-2.0 licence.

Summary of the jjcoders/suna-v1 model card, 2026-10-22

What it is

What it runs on

Memory and cards for suna-v1 (FP16)

1.1 GBweights, file size

1.6 GBruntime overhead

From the model card

Read the model card

Suna AI v1.0 by JJ coders Group (suna.jjcoders.com) is a 3.0-Billion Parameter AI Music Production & Vocal Co-Producer System:

🎙️ Sing-to-Masterpiece Vocal Co-Producer: Analyzes raw human singing/humming for key & tempo, applies studio vocal polish, generates a key/BPM-locked multi-instrument band + cinematic sound effects, and sidechain-mixes the vocal and band in 44.1 kHz stereo.
🌐 Brain 1 (JJ_coders_Suna_AI_v1_Q4_K_M.gguf — 1.12 GB): Qwen2.5-1.5B-Instruct Open-World Musicologist across 10,000+ global genres.
🎵 Brain 2 (model.safetensors — 4.07 GB): MusicGen-Medium 1.5B Audio Foundation Synthesis Engine in FP16.

Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.

Details