Skip to content

Malaysian TTS 0.6B V1 VRAM requirements

How much VRAM Malaysian TTS 0.6B V1 needs
4-bit weights: 364 MBWith 8K context: 2.1 GBWith 32K context: 4.7 GB8-bit, 8K context: 2.3 GB
Parameters
629M
Longest context
32K
Released
Aug 2025
Strongest at
Not rated

How much VRAM Malaysian TTS 0.6B V1 needs

Memory Malaysian TTS 0.6B V1 needs at each precision and context length
PrecisionWeightsWith 8KWith 32K
16-bit1.2 GB2.9 GB5.5 GB
8-bit638 MB2.3 GB5.0 GB
6-bit492 MB2.2 GB4.8 GB
5-bit427 MB2.1 GB4.7 GB
4-bit364 MB2.1 GB4.7 GB
3-bit293 MB2.0 GB4.6 GB
2-bit225 MB1.9 GB4.6 GB
1.6-bit124 MB1.8 GB4.5 GB

One conversation, 16-bit context cache, runtime overhead included.

GPUs and Macs that run Malaysian TTS 0.6B V1

The best way Malaysian TTS 0.6B V1 runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40608-bit32K230 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4060 report
RTX 50608-bit32K336 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 5060 report
RTX 30708-bit32K336 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 3070 report
RTX 3080 10 GB8-bit32K477 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 3080 10 GB report
RTX 3060 12 GB8-bit32K286 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 3060 12 GB report
Arc B5808-bit32K217 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Arc B580 report
RTX 40708-bit32K366 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4070 report
RTX 4070 Super8-bit32K366 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4070 Super report
RTX 50708-bit32K442 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 5070 report
RTX 4060 Ti 16 GB8-bit32K241 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit32K336 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit32K246 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Arc A770 16 GB report
Radeon RX 9070 XT8-bit32K318 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit32K442 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4070 Ti Super report
RTX 4080 Super8-bit32K468 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4080 Super report
RTX 5070 Ti8-bit32K525 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 5070 Ti report
RTX 50808-bit32K545 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 5080 report
Radeon RX 7900 XTX8-bit32K390 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Radeon RX 7900 XTX report
RTX 30908-bit32K537 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 3090 report
RTX 40908-bit32K559 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 4090 report
Radeon AI PRO R97008-bit32K318 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Radeon AI PRO R9700 report
RTX 50908-bit32K728 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 5090 report
2× RTX 30908-bit32K537 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 2× RTX 3090 report
2× RTX 40908-bit32K559 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit32K645 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit32K545 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX 6000 Ada report
L40S8-bit32K514 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on L40S report
2× RTX 50908-bit32K728 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 2× RTX 5090 report
4× RTX 30908-bit32K537 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit32K728 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB8-bit32K111 tok/s0.7 sPDF: Malaysian TTS 0.6B V1 on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit32K111 tok/s0.7 sPDF: Malaysian TTS 0.6B V1 on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit32K111 tok/s0.7 sPDF: Malaysian TTS 0.6B V1 on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit32K191 tok/s0.3 sPDF: Malaysian TTS 0.6B V1 on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit32K191 tok/s0.3 sPDF: Malaysian TTS 0.6B V1 on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit32K191 tok/s0.3 sPDF: Malaysian TTS 0.6B V1 on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit32K258 tok/s0.2 sPDF: Malaysian TTS 0.6B V1 on M4 Max 128 GB report
M5 Max 128 GB8-bit32K271 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on M5 Max 128 GB report
M3 Ultra 96 GB8-bit32K286 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit32K286 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit32K286 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit32K323 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit32K174 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit32K174 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit32K248 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on RTX Spark report
DGX Spark8-bit32K230 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on DGX Spark report
Dual DGX Spark8-bit32K230 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on Dual DGX Spark report
DGX Station GB3008-bit32K1,026 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit32K728 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit32K728 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit32K728 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit32K764 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on A100 80 GB SXM report
H100 SXM8-bit32K889 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on H100 SXM report
4× H200 NVL8-bit32K962 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 4× H200 NVL report
8× H200 NVL8-bit32K962 tok/s0.1 sPDF: Malaysian TTS 0.6B V1 on 8× H200 NVL report

Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.

How to run it

llama.cpp
llama-server -hf mradermacher/Malaysian-TTS-0.6B-v1-GGUF
vLLM
vllm serve mesolitica/Malaysian-TTS-0.6B-v1

Questions

How much VRAM does Malaysian TTS 0.6B V1 need?

Malaysian TTS 0.6B V1 needs 364 MB for its weights at 4-bit, 2.1 GB with an 8K context and 4.7 GB with 32K. At 8-bit, 2.3 GB.

What is the minimum VRAM for Malaysian TTS 0.6B V1?

About 2.0 GB at 3-bit with an 8K context, the lowest precision we recommend for a model this size. The smallest machine here that runs it without spilling is the RTX 4060.

How big is Malaysian TTS 0.6B V1?

629M parameters: 1.2 GB of weights at 16-bit, 364 MB at 4-bit.

Can Malaysian TTS 0.6B V1 run on an RTX 4090?

Yes. At 8-bit it holds 32K of context and writes about 559 tokens a second.

More models