What hardware does Veena run on?

Tuned builds: NVIDIA H100 SXM5. plowrt also runs them on H100 NVL, H200 SXM.

Does Veena fit on one GPU?

Yes. The sm90a-h100-tp1 build runs on one NVIDIA H100 SXM5.

Can I run Veena on AMD?

Not with a tuned build yet. Today's builds are for NVIDIA.

What context length does Veena support?

The upstream model supports 131,072 tokens. Our builds serve up to 2,048.

What license is Veena under?

Apache-2.0. Check the checkpoint's license on Hugging Face before commercial use.

Install plowrt, then run plowrt load veena --fetch-weights and plowrt serve --model veena --port 8080. It serves an OpenAI-compatible API.

Text to speech 3B 128k native context bf16 Apache-2.0

Veena

Maya Research's Veena, a Llama-style codec language model with a SNAC decoder for Hindi and English speech.

maya-research/Veena

Specs

Developer: Maya Research
Parameters: 3B
Task: Text to speech
Context: 131,072 tokens upstream, 2,048 served by our builds
Precisions: bf16
License: Apache-2.0
Checkpoint: maya-research/Veena
Updated: 28 Oct 2026

Where it runs Each build is tuned for one part. plowrt runs it beyond that.

Tuned for

NVIDIA H100 SXM5

Runs as-is on H100 NVL, H200 SXM

Elsewhere plowrt still runs the model on other NVIDIA, AMD and Apple hardware, without this build’s tuning: expect lower throughput.

Run it An OpenAI-compatible server on your own machine.

plowrt 0.1.3 Install, pull, serve, call

curl -fsSL https://pdist.infervisor.ai/install.sh | sh

Install

```
plowrt load veena --fetch-weights
```
Pull with weights
```
plowrt serve --model veena --port 8080
```
Serve

curl -s http://127.0.0.1:8080/v1/audio/speech \
  -H 'content-type: application/json' \
  -d '{"model":"veena","input":"Hello from plow.","voice":"kavya"}' -o out.wav

Try it

Builds `plowrt load` picks the closest fit for your machine.

NVIDIA H100 SXM5 80 GiB, also H100 NVL, H200 SXM bf16 1 GPU 2k Published

plowrt load infervisor/veena:sm90a-h100-tp1 --fetch-weights

This build only

Target: sm90a-h100-tp1 , tuned on NVIDIA H100 SXM5; runs as-is on H100 NVL, H200 SXM; other hardware runs it untuned
Checkpoint: maya-research/Veena
Published: 28 Oct 2026
Compiled by: plowc 0.1.3 , from this recipe

About

Veena predicts SNAC audio codes with a 3B Llama-architecture LM; the codec is lowered to its own packet (codec.pkt) and travels in the bundle. Speech is served through /v1/audio/speech.

Questions

What hardware does Veena run on?: Tuned builds: NVIDIA H100 SXM5. plowrt also runs them on H100 NVL, H200 SXM.
Does Veena fit on one GPU?: Yes. The sm90a-h100-tp1 build runs on one NVIDIA H100 SXM5.
Can I run Veena on AMD?: Not with a tuned build yet. Today's builds are for NVIDIA.
What context length does Veena support?: The upstream model supports 131,072 tokens. Our builds serve up to 2,048.
What license is Veena under?: Apache-2.0. Check the checkpoint's license on Hugging Face before commercial use.
How do I run Veena?: Install plowrt, then run plowrt load veena --fetch-weights and plowrt serve --model veena --port 8080. It serves an OpenAI-compatible API.