audio model · dia · Windows
DI Can I run Dia 1.6B on Nvidia GeForce RTX 3090 (24GB)?
Yes. Dia 1.6B runs on Nvidia GeForce RTX 3090 (24GB) at fp16 (~10 GB of ~23 GB usable).
Runs at fp16 using ~10 GB of ~23 GB usable.
- Peak memory
- ~10 GB
- Usable on device
- ~23 GB
- Device memory
- 24 GB
- Quant
- fp16
How to run it
Use PyTorch (CUDA) or nari-labs/dia at fp16. It loads as a single model and runs on a GPU.
- Type
- Text to speech (dialogue)
- Parameters
- 1.6B
- Peak memory
- ~10 GB at fp16
- License
- Apache-2.0
- Memory
- 24 GB vram
- Usable for weights
- ~23 GB
- Best runtime
- vLLM (Linux) / Ollama (CUDA)
You could also run
Run Dia 1.6B on other hardware
FAQ
Can Nvidia GeForce RTX 3090 (24GB) run Dia 1.6B?
Yes. Dia 1.6B runs on Nvidia GeForce RTX 3090 (24GB) at fp16 (~10 GB of ~23 GB usable).
How much memory does Dia 1.6B need?
Nvidia GeForce RTX 3090 (24GB) has room to spare. At fp16 the realistic peak is ~10 GB of memory.
What do I use to run Dia 1.6B locally?
Dia 1.6B runs in PyTorch (CUDA) or nari-labs/dia. It needs a GPU.
Sources
VRAM figures are sourced peak-usage anchors at the noted quant, validated 2026-06-15. See methodology.