Moonshine Streaming Tiny needs about 0.6 GB of memory at F16 (a 0.1 GB file) with an 8K context window, so it fits on a 4 GB GPU. This repository has 3 quants, from 0.1 GB (Q8_0) to 0.2 GB (F32).
Moonshine Streaming Tiny – lightweight ASR for edge devices
Moonshine Streaming Tiny is a sequence‑to‑sequence ASR model with a sliding‑window Transformer encoder and autoregressive decoder. It is trained for low‑latency English speech transcription on memory‑constrained hardware.
Good at
- Low‑latency streaming ASR
- Edge‑class hardware support
- English‑only transcription
Limits
- No multilingual support
- No vision or timestamp output
- Not for high‑stakes or surveillance use
Best for
- Live captioning on smartphones
- Voice command systems
- Real‑time transcription in embedded devices
Who it’s for. Developers building on‑device speech interfaces that require sub‑1 GB memory and low compute. The model is suitable for applications needing quick, on‑device transcription without external servers.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q8_0 | 0.1 GB | 0.6 GB | Fits | Get file → | |
| F16 | 0.1 GB | 0.6 GB | Fits | Get file → | |
| F32 | 0.2 GB | 0.7 GB | Fits | Get file → |
For 8 GB VRAM, Q8_0 (0.05 GB) is ideal, using ~0.6 GB at 8 k context. If 4 GB VRAM is available, Q8_0 remains safe. For 2 GB VRAM, Q8_0 still fits, but memory usage is close to the limit. F16 and F32 require more VRAM and are not recommended for low‑budget devices.
What GPU do you need for Moonshine Streaming Tiny?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is F32 (0.2 GB file, about 0.7 GB in use).
Which Moonshine Streaming Tiny quant should you choose?
F16 (0.1 GB) is the middle option in this repository. The unquantized model. It gives original quality and needs the most memory, so it is mainly for conversion or reference.
If memory is tight, the smallest file is Q8_0 at 0.1 GB, with the quality trade-offs that come with fewer bits. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Moonshine Streaming Tiny
How much VRAM does Moonshine Streaming Tiny need?
About 0.6 GB at F16 with an 8K context window. That is the 0.1 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Moonshine Streaming Tiny on an 8 GB GPU?
Yes, with F32: the 0.2 GB file needs about 0.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Moonshine Streaming Tiny GGUF?
It depends on the quant: Q8_0 is 0.1 GB, F16 is 0.1 GB, F32 is 0.2 GB.
Where can I download Moonshine Streaming Tiny GGUF?
The files are on Hugging Face in the handy-computer/moonshine-streaming-tiny-gguf repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Moonshine Streaming Tiny use?
Hugging Face lists the license as mit. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Moonshine Streaming Tiny on 4 GB VRAM?
Yes, the Q8_0 quantized model uses about 0.6 GB at 8 k context, well within a 4 GB VRAM budget.
Does this model support multiple languages?
No, it is trained and evaluated only on English speech.
Is there a timestamp output?
No, the model produces plain text transcripts without timestamps.