Skip to content

Yi 6B 200K VRAM requirements

How much VRAM Yi 6B 200K needs
4-bit weights: 3.4 GBWith 8K context: 4.6 GBWith 32K context: 6.1 GB8-bit, 8K context: 7.2 GB
Parameters
6.1B
Longest context
195K
Released
Nov 2023
License
Apache 2.0
Strongest at
Not rated

How much VRAM Yi 6B 200K needs

Memory Yi 6B 200K needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit11.3 GB12.5 GB14.0 GB20.0 GB
8-bit6.0 GB7.2 GB8.7 GB14.7 GB
6-bit4.6 GB5.8 GB7.3 GB13.3 GB
5-bit4.0 GB5.2 GB6.7 GB12.7 GB
4-bit3.4 GB4.6 GB6.1 GB12.1 GB
3-bit2.8 GB4.0 GB5.5 GB11.5 GB
2-bit2.1 GB3.3 GB4.8 GB10.8 GB
1.6-bit1.2 GB2.4 GB3.9 GB9.9 GB

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

GPUs and Macs that run Yi 6B 200K

The best way Yi 6B 200K runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40606-bit36K39 tok/s0.8 sPDF: Yi 6B 200K on RTX 4060 report
RTX 50606-bit36K62 tok/s0.7 sPDF: Yi 6B 200K on RTX 5060 report
RTX 30706-bit36K62 tok/s0.6 sPDF: Yi 6B 200K on RTX 3070 report
RTX 3080 10 GB8-bit44K80 tok/s0.4 sPDF: Yi 6B 200K on RTX 3080 10 GB report
RTX 3060 12 GB8-bit72K39 tok/s1.0 sPDF: Yi 6B 200K on RTX 3060 12 GB report
Arc B5808-bit72K35 tok/s0.7 sPDF: Yi 6B 200K on Arc B580 report
RTX 40708-bit72K54 tok/s0.4 sPDF: Yi 6B 200K on RTX 4070 report
RTX 4070 Super8-bit72K54 tok/s0.4 sPDF: Yi 6B 200K on RTX 4070 Super report
RTX 50708-bit72K71 tok/s0.4 sPDF: Yi 6B 200K on RTX 5070 report
RTX 4060 Ti 16 GB8-bit136K32 tok/s0.6 sPDF: Yi 6B 200K on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit136K49 tok/s0.5 sPDF: Yi 6B 200K on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit136K42 tok/s0.6 sPDF: Yi 6B 200K on Arc A770 16 GB report
Radeon RX 9070 XT8-bit136K57 tok/s0.3 sPDF: Yi 6B 200K on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit136K71 tok/s0.3 sPDF: Yi 6B 200K on RTX 4070 Ti Super report
RTX 4080 Super8-bit136K77 tok/s0.2 sPDF: Yi 6B 200K on RTX 4080 Super report
RTX 5070 Ti8-bit136K93 tok/s0.3 sPDF: Yi 6B 200K on RTX 5070 Ti report
RTX 50808-bit136K99 tok/s0.2 sPDF: Yi 6B 200K on RTX 5080 report
Radeon RX 7900 XTX8-bit195K81 tok/s0.6 sPDF: Yi 6B 200K on Radeon RX 7900 XTX report
RTX 30908-bit195K96 tok/s0.4 sPDF: Yi 6B 200K on RTX 3090 report
RTX 40908-bit195K103 tok/s0.2 sPDF: Yi 6B 200K on RTX 4090 report
Radeon AI PRO R97008-bit195K57 tok/s0.4 sPDF: Yi 6B 200K on Radeon AI PRO R9700 report
RTX 50908-bit195K170 tok/s0.1 sPDF: Yi 6B 200K on RTX 5090 report
2× RTX 30908-bit195K96 tok/s0.2 sPDF: Yi 6B 200K on 2× RTX 3090 report
2× RTX 40908-bit195K103 tok/s0.1 sPDF: Yi 6B 200K on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit195K133 tok/s0.2 sPDF: Yi 6B 200K on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit195K99 tok/s0.1 sPDF: Yi 6B 200K on RTX 6000 Ada report
L40S8-bit195K90 tok/s0.2 sPDF: Yi 6B 200K on L40S report
2× RTX 50908-bit195K170 tok/s0.1 sPDF: Yi 6B 200K on 2× RTX 5090 report
4× RTX 30908-bit195K96 tok/s0.1 sPDF: Yi 6B 200K on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit195K170 tok/s0.1 sPDF: Yi 6B 200K on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB8-bit56K15 tok/s8.0 sPDF: Yi 6B 200K on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit136K15 tok/s8.0 sPDF: Yi 6B 200K on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit195K15 tok/s8.0 sPDF: Yi 6B 200K on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit136K33 tok/s4.0 sPDF: Yi 6B 200K on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit195K33 tok/s4.0 sPDF: Yi 6B 200K on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit195K33 tok/s4.0 sPDF: Yi 6B 200K on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit195K57 tok/s2.0 sPDF: Yi 6B 200K on M4 Max 128 GB report
M5 Max 128 GB8-bit195K63 tok/s0.6 sPDF: Yi 6B 200K on M5 Max 128 GB report
M3 Ultra 96 GB8-bit195K71 tok/s1.1 sPDF: Yi 6B 200K on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit195K71 tok/s1.1 sPDF: Yi 6B 200K on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit195K71 tok/s1.1 sPDF: Yi 6B 200K on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit195K96 tok/s0.3 sPDF: Yi 6B 200K on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit195K24 tok/s1.1 sPDF: Yi 6B 200K on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit195K24 tok/s1.1 sPDF: Yi 6B 200K on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit195K33 tok/s0.4 sPDF: Yi 6B 200K on RTX Spark report
DGX Spark8-bit195K30 tok/s0.4 sPDF: Yi 6B 200K on DGX Spark report
Dual DGX Spark8-bit195K30 tok/s0.2 sPDF: Yi 6B 200K on Dual DGX Spark report
DGX Station GB3008-bit195K455 tok/s0.1 sPDF: Yi 6B 200K on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit195K170 tok/s0.1 sPDF: Yi 6B 200K on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit195K170 tok/s0.1 sPDF: Yi 6B 200K on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit195K170 tok/s0.1 sPDF: Yi 6B 200K on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit195K190 tok/s0.3 sPDF: Yi 6B 200K on A100 80 GB SXM report
H100 SXM8-bit195K279 tok/s0.1 sPDF: Yi 6B 200K on H100 SXM report
4× H200 NVL8-bit195K358 tok/s0.1 sPDF: Yi 6B 200K on 4× H200 NVL report
8× H200 NVL8-bit195K358 tok/s0.1 sPDF: Yi 6B 200K 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 TheBloke/Yi-6B-200K-GGUF
vLLM
vllm serve 01-ai/Yi-6B-200K

Questions

How much VRAM does Yi 6B 200K need?

Yi 6B 200K needs 3.4 GB for its weights at 4-bit, 4.6 GB with an 8K context and 6.1 GB with 32K. At 8-bit, 7.2 GB.

What is the minimum VRAM for Yi 6B 200K?

About 4.0 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 Yi 6B 200K?

6.1B parameters: 11.3 GB of weights at 16-bit, 3.4 GB at 4-bit.

Can Yi 6B 200K run on an RTX 4090?

Yes. At 8-bit it holds 195K of context and writes about 103 tokens a second.

More models