How much VRAM does Fun-CosyVoice3-0.5B-2512 need?

Q8_0 is exactly 360,751,936 bytes (0.34 GiB) in weights. Add the KV cache, which depends on context length, plus roughly half a gigabyte of runtime overhead.

FunAudioLLM · audio tts

Fun-CosyVoice3-0.5B-2512

FunAudioLLM/Fun-CosyVoice3-0.5B-2512

Fun-CosyVoice3-0.5B-2512 at Q8_0 is exactly 360,751,936 bytes (0.34 GiB / 0.36 GB)

From the file· summed from 1 file(s)From the file· KV per layer

Parameters

—

Architecture

cosyvoice3-campplus

null layers

Context

—

native (config.json)

License

apache-2.0

Shipped quantizations

● exact bytes, summed from published files

Quant	Size●	Exact bytes●	Effective bpw●	Tensors●	Publisher
Q8_0	0.34 GiB	360,751,936	—	—	cstr
Q4_K3 shards	0.85 GiB	913,040,640	—	—	cstr
F166 shards	3.52 GiB	3,784,611,328	—	—	cstr

Compare with

same modality, comparable size

vs Ace-Step1.5 vs Qwen3-TTS-12Hz-0.6B-Base vs OmniVoice vs VieNeu-TTS-0.3B vs neutts-air vs chatterbox All alternatives →Will it fit on my card? →

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q8_0 at roughly 0.00 GiB. The real file is 0.34 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json

Layers

—

Attention heads

—

KV heads

—

Head dim

—

Hidden size

—

Vocab

—

Sliding window

none

SWA period

—

MLA

Experts

—

Experts per token

—

use_sliding_window

—

Questions people ask

How much VRAM does Fun-CosyVoice3-0.5B-2512 need?: Q8_0 is exactly 360,751,936 bytes (0.34 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of Fun-CosyVoice3-0.5B-2512 should I use?: Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.