Skip to content

Yi Coder 9B VRAM requirements

How much VRAM Yi Coder 9B needs
4-bit weights: 5.0 GBWith 8K context: 6.4 GBWith 32K context: 8.7 GB8-bit, 8K context: 10.2 GB
Parameters
8.8B
Longest context
128K
Released
Aug 2024
License
Apache 2.0
Strongest at
Coding

How much VRAM Yi Coder 9B needs

Memory Yi Coder 9B needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit16.4 GB17.9 GB20.1 GB29.1 GB
8-bit8.7 GB10.2 GB12.4 GB21.4 GB
6-bit6.7 GB8.2 GB10.4 GB19.4 GB
5-bit5.8 GB7.3 GB9.5 GB18.5 GB
4-bit5.0 GB6.4 GB8.7 GB17.7 GB
3-bit4.0 GB5.5 GB7.7 GB16.7 GB
2-bit3.1 GB4.5 GB6.8 GB15.8 GB
1.6-bit1.7 GB3.1 GB5.4 GB14.4 GB

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

GPUs and Macs that run Yi Coder 9B

The best way Yi Coder 9B runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40604-bit20K35 tok/s1.2 sPDF: Yi Coder 9B on RTX 4060 report
RTX 50604-bit20K56 tok/s1.0 sPDF: Yi Coder 9B on RTX 5060 report
RTX 30704-bit20K56 tok/s0.9 sPDF: Yi Coder 9B on RTX 3070 report
RTX 3080 10 GB6-bit20K68 tok/s0.6 sPDF: Yi Coder 9B on RTX 3080 10 GB report
RTX 3060 12 GB8-bit20K27 tok/s1.5 sPDF: Yi Coder 9B on RTX 3060 12 GB report
Arc B5808-bit20K23 tok/s1.0 sPDF: Yi Coder 9B on Arc B580 report
RTX 40708-bit20K37 tok/s0.6 sPDF: Yi Coder 9B on RTX 4070 report
RTX 4070 Super8-bit20K37 tok/s0.5 sPDF: Yi Coder 9B on RTX 4070 Super report
RTX 50708-bit20K48 tok/s0.6 sPDF: Yi Coder 9B on RTX 5070 report
RTX 4060 Ti 16 GB8-bit60K21 tok/s0.9 sPDF: Yi Coder 9B on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit60K33 tok/s0.8 sPDF: Yi Coder 9B on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit60K28 tok/s0.9 sPDF: Yi Coder 9B on Arc A770 16 GB report
Radeon RX 9070 XT8-bit60K38 tok/s0.5 sPDF: Yi Coder 9B on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit60K48 tok/s0.4 sPDF: Yi Coder 9B on RTX 4070 Ti Super report
RTX 4080 Super8-bit60K52 tok/s0.4 sPDF: Yi Coder 9B on RTX 4080 Super report
RTX 5070 Ti8-bit60K63 tok/s0.4 sPDF: Yi Coder 9B on RTX 5070 Ti report
RTX 50808-bit60K67 tok/s0.3 sPDF: Yi Coder 9B on RTX 5080 report
Radeon RX 7900 XTX8-bit128K55 tok/s0.8 sPDF: Yi Coder 9B on Radeon RX 7900 XTX report
RTX 30908-bit128K65 tok/s0.5 sPDF: Yi Coder 9B on RTX 3090 report
RTX 40908-bit128K70 tok/s0.2 sPDF: Yi Coder 9B on RTX 4090 report
Radeon AI PRO R97008-bit128K38 tok/s0.5 sPDF: Yi Coder 9B on Radeon AI PRO R9700 report
RTX 50908-bit128K115 tok/s0.2 sPDF: Yi Coder 9B on RTX 5090 report
2× RTX 30908-bit128K65 tok/s0.3 sPDF: Yi Coder 9B on 2× RTX 3090 report
2× RTX 40908-bit128K70 tok/s0.1 sPDF: Yi Coder 9B on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit128K90 tok/s0.3 sPDF: Yi Coder 9B on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit128K67 tok/s0.2 sPDF: Yi Coder 9B on RTX 6000 Ada report
L40S8-bit128K61 tok/s0.3 sPDF: Yi Coder 9B on L40S report
2× RTX 50908-bit128K115 tok/s0.1 sPDF: Yi Coder 9B on 2× RTX 5090 report
4× RTX 30908-bit128K65 tok/s0.2 sPDF: Yi Coder 9B on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit128K115 tok/s0.1 sPDF: Yi Coder 9B on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB6-bit28K13 tok/s12 sPDF: Yi Coder 9B on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit60K10 tok/s12 sPDF: Yi Coder 9B on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit112K10 tok/s12 sPDF: Yi Coder 9B on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit60K22 tok/s6.0 sPDF: Yi Coder 9B on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit128K22 tok/s6.0 sPDF: Yi Coder 9B on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit128K22 tok/s6.0 sPDF: Yi Coder 9B on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit128K38 tok/s3.0 sPDF: Yi Coder 9B on M4 Max 128 GB report
M5 Max 128 GB8-bit128K42 tok/s0.9 sPDF: Yi Coder 9B on M5 Max 128 GB report
M3 Ultra 96 GB8-bit128K48 tok/s1.7 sPDF: Yi Coder 9B on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit128K48 tok/s1.7 sPDF: Yi Coder 9B on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit128K48 tok/s1.7 sPDF: Yi Coder 9B on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit128K64 tok/s0.5 sPDF: Yi Coder 9B on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit128K16 tok/s1.7 sPDF: Yi Coder 9B on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit128K16 tok/s1.7 sPDF: Yi Coder 9B on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit128K22 tok/s0.6 sPDF: Yi Coder 9B on RTX Spark report
DGX Spark8-bit128K20 tok/s0.6 sPDF: Yi Coder 9B on DGX Spark report
Dual DGX Spark8-bit128K20 tok/s0.3 sPDF: Yi Coder 9B on Dual DGX Spark report
DGX Station GB3008-bit128K306 tok/s0.1 sPDF: Yi Coder 9B on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit128K115 tok/s0.1 sPDF: Yi Coder 9B on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit128K115 tok/s0.1 sPDF: Yi Coder 9B on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit128K115 tok/s0.1 sPDF: Yi Coder 9B on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit128K128 tok/s0.4 sPDF: Yi Coder 9B on A100 80 GB SXM report
H100 SXM8-bit128K188 tok/s0.1 sPDF: Yi Coder 9B on H100 SXM report
4× H200 NVL8-bit128K241 tok/s0.1 sPDF: Yi Coder 9B on 4× H200 NVL report
8× H200 NVL8-bit128K241 tok/s0.1 sPDF: Yi Coder 9B 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/Yi-Coder-9B-i1-GGUF
vLLM
vllm serve 01-ai/Yi-Coder-9B

Questions

How much VRAM does Yi Coder 9B need?

Yi Coder 9B needs 5.0 GB for its weights at 4-bit, 6.4 GB with an 8K context and 8.7 GB with 32K. At 8-bit, 10.2 GB.

What is the minimum VRAM for Yi Coder 9B?

About 5.5 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 Yi Coder 9B?

8.8B parameters: 16.4 GB of weights at 16-bit, 5.0 GB at 4-bit.

Can Yi Coder 9B run on an RTX 4090?

Yes. At 8-bit it holds 128K of context and writes about 70 tokens a second.

More models