Skip to content

Yi 34B Chat VRAM requirements

How much VRAM Yi 34B Chat needs
4-bit weights: 19.4 GBWith 4K context: 21.1 GB8-bit, 4K context: 35.8 GB
Parameters
34.4B
Longest context
4K
Released
Nov 2023
License
Apache 2.0
Strongest at
Not rated

How much VRAM Yi 34B Chat needs

Memory Yi 34B Chat needs at each precision and context length
PrecisionWeightsWith 4K
16-bit64.1 GB65.8 GB
8-bit34.0 GB35.8 GB
6-bit26.3 GB28.0 GB
5-bit22.8 GB24.5 GB
4-bit19.4 GB21.1 GB
3-bit15.7 GB17.4 GB
2-bit12.0 GB13.7 GB
1.6-bit6.6 GB8.3 GB

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

GPUs and Macs that run Yi 34B Chat

The best way Yi 34B Chat runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: Yi 34B Chat on RTX 4060 report
RTX 5060Too bigPDF: Yi 34B Chat on RTX 5060 report
RTX 3070Too bigPDF: Yi 34B Chat on RTX 3070 report
RTX 3080 10 GBToo bigPDF: Yi 34B Chat on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: Yi 34B Chat on RTX 3060 12 GB report
Arc B580Too bigPDF: Yi 34B Chat on Arc B580 report
RTX 4070Too bigPDF: Yi 34B Chat on RTX 4070 report
RTX 4070 SuperToo bigPDF: Yi 34B Chat on RTX 4070 Super report
RTX 5070Too bigPDF: Yi 34B Chat on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: Yi 34B Chat on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: Yi 34B Chat on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: Yi 34B Chat on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: Yi 34B Chat on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: Yi 34B Chat on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: Yi 34B Chat on RTX 4080 Super report
RTX 5070 TiToo bigPDF: Yi 34B Chat on RTX 5070 Ti report
RTX 5080Too bigPDF: Yi 34B Chat on RTX 5080 report
Radeon RX 7900 XTX4-bit4K26 tok/s3.3 sPDF: Yi 34B Chat on Radeon RX 7900 XTX report
RTX 30904-bit4K30 tok/s2.1 sPDF: Yi 34B Chat on RTX 3090 report
RTX 40904-bit4K32 tok/s0.9 sPDF: Yi 34B Chat on RTX 4090 report
Radeon AI PRO R97006-bit4K13 tok/s2.1 sPDF: Yi 34B Chat on Radeon AI PRO R9700 report
RTX 50906-bit4K42 tok/s0.7 sPDF: Yi 34B Chat on RTX 5090 report
2× RTX 30908-bit4K18 tok/s1.2 sPDF: Yi 34B Chat on 2× RTX 3090 report
2× RTX 40908-bit4K19 tok/s0.5 sPDF: Yi 34B Chat on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit4K25 tok/s1.1 sPDF: Yi 34B Chat on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit4K18 tok/s0.8 sPDF: Yi 34B Chat on RTX 6000 Ada report
L40S8-bit4K16 tok/s1.3 sPDF: Yi 34B Chat on L40S report
2× RTX 50908-bit4K33 tok/s0.4 sPDF: Yi 34B Chat on 2× RTX 5090 report
4× RTX 30908-bit4K18 tok/s0.6 sPDF: Yi 34B Chat on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit4K33 tok/s0.6 sPDF: Yi 34B Chat on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: Yi 34B Chat on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: Yi 34B Chat on Mac mini M4 24 GB report
Mac mini M4 32 GB4-bit4K4.7 tok/s47 sPDF: Yi 34B Chat on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: Yi 34B Chat on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit4K6.0 tok/s24 sPDF: Yi 34B Chat on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit4K6.0 tok/s24 sPDF: Yi 34B Chat on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit4K11 tok/s12 sPDF: Yi 34B Chat on M4 Max 128 GB report
M5 Max 128 GB8-bit4K12 tok/s3.4 sPDF: Yi 34B Chat on M5 Max 128 GB report
M3 Ultra 96 GB8-bit4K14 tok/s6.6 sPDF: Yi 34B Chat on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit4K14 tok/s6.6 sPDF: Yi 34B Chat on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit4K14 tok/s6.6 sPDF: Yi 34B Chat on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit4K20 tok/s2.1 sPDF: Yi 34B Chat on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit4K4.2 tok/s6.8 sPDF: Yi 34B Chat on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit4K4.2 tok/s6.8 sPDF: Yi 34B Chat on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit4K5.8 tok/s2.4 sPDF: Yi 34B Chat on RTX Spark report
DGX Spark8-bit4K5.3 tok/s2.4 sPDF: Yi 34B Chat on DGX Spark report
Dual DGX Spark8-bit4K5.3 tok/s1.4 sPDF: Yi 34B Chat on Dual DGX Spark report
DGX Station GB3008-bit4K110 tok/s0.2 sPDF: Yi 34B Chat on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit4K33 tok/s0.3 sPDF: Yi 34B Chat on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit4K33 tok/s0.2 sPDF: Yi 34B Chat on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit4K33 tok/s0.1 sPDF: Yi 34B Chat on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit4K37 tok/s1.5 sPDF: Yi 34B Chat on A100 80 GB SXM report
H100 SXM8-bit4K58 tok/s0.5 sPDF: Yi 34B Chat on H100 SXM report
4× H200 NVL8-bit4K80 tok/s0.2 sPDF: Yi 34B Chat on 4× H200 NVL report
8× H200 NVL8-bit4K80 tok/s0.1 sPDF: Yi 34B Chat 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-34B-Chat-GGUF
vLLM
vllm serve 01-ai/Yi-34B-Chat

Questions

How much VRAM does Yi 34B Chat need?

Yi 34B Chat needs 19.4 GB for its weights at 4-bit, 21.1 GB with an 4K context and 21.1 GB with 4K. At 8-bit, 35.8 GB.

What is the minimum VRAM for Yi 34B Chat?

About 17.4 GB at 3-bit with an 4K context, the lowest precision we recommend for a model this size. The smallest machine here that runs it without spilling is the Radeon RX 7900 XTX.

How big is Yi 34B Chat?

34.4B parameters: 64.1 GB of weights at 16-bit, 19.4 GB at 4-bit.

Can Yi 34B Chat run on an RTX 4090?

Yes. At 4-bit it holds 4K of context and writes about 32 tokens a second.

More models