drama-3 needs about 26.7 GB of memory at Q4_K_S (a 23.4 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 6 quants, from 17.1 GB (Q2_K) to 25.6 GB (F32).
Drama 3 GGUF – audio‑centric text‑to‑speech model
Drama 3 is a text‑to‑speech model built on Fish Audio’s architecture. It supports multi‑speaker generation, voice cloning, and natural‑language control. The model is available in several GGUF quantizations for llama.cpp.
Good at
- Multi‑speaker TTS
- Voice cloning
- Long‑form audio generation
Limits
- No vision support
- Large VRAM requirement for full precision
- License not specified
Best for
- Audio production pipelines
- Voice‑over creation
- Interactive storytelling
Who it’s for. Engineers and content creators who need high‑quality, multi‑character speech synthesis. The model is suitable for developers using llama.cpp or audio.cpp for local inference.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 17.1 GB | 19.6 GB | Too big | Get file → | |
| Q3_K_S | 20.3 GB | 23.3 GB | Too big | Get file → | |
| Q4_K_S | 23.4 GB | 26.7 GB | Too big | Get file → | |
| Q4_K_S | 24.6 GB | 28.0 GB | Too big | Get file → | |
| F32 | 24.7 GB | 28.1 GB | Too big | Get file → | |
| F32 | 25.6 GB | 29.2 GB | Too big | Get file → |
For 8 GB VRAM use Q2_K (17 GB file, ~20 GB memory). For 12 GB VRAM use Q3_K_S (20 GB file, ~23 GB memory). For 16 GB VRAM use Q4_K_S (23 GB file, ~27 GB memory). Full precision F32 requires 24–26 GB file and ~28 GB memory.
What GPU do you need for drama-3?
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: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q3_K_S (20.3 GB file, about 23.3 GB in use).
- 32 GB VRAM: the best quant that fits is F32 (25.6 GB file, about 29.2 GB in use).
Which drama-3 quant should you choose?
Q4_K_S (23.4 GB) is the middle option in this repository. The usual sweet spot between file size and quality, and the quant most people start with.
If memory is tight, the smallest file is Q2_K at 17.1 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about drama-3
How much VRAM does drama-3 need?
About 26.7 GB at Q4_K_S with an 8K context window. That is the 23.4 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 drama-3 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 19.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of drama-3 GGUF?
It depends on the quant: Q2_K is 17.1 GB, Q3_K_S is 20.3 GB, Q4_K_S is 23.4 GB, Q4_K_S is 24.6 GB, F32 is 24.7 GB, F32 is 25.6 GB.
Where can I download drama-3 GGUF?
The files are on Hugging Face in the Jmica/Drama-3-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 drama-3 use?
Hugging Face lists the license as other. 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 Drama 3 on 8 GB VRAM?
Use the Q2_K quantization; it requires about 20 GB of VRAM at 8 k context, so 8 GB is insufficient.
Does Drama 3 support image inputs?
No, the model has no vision capability.
What is the context length?
The model is trained for 8 k tokens, with memory estimates provided for that context.