Skip to content

Laguna S 2.1 VRAM requirements

How much VRAM Laguna S 2.1 needs
4-bit weights: 66.4 GBWith 8K context: 67.6 GBWith 32K context: 68.7 GB8-bit, 8K context: 118 GB
Parameters
118B
Active per token
8.4B
Longest context
1M
Released
Jul 2026
License
openmdw-1.1
Strongest at
Not rated

Its licence (openmdw-1.1) is not a standard open-source one: check the terms before commercial use.

How much VRAM Laguna S 2.1 needs

Memory Laguna S 2.1 needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit219 GB220 GB221 GB226 GB
8-bit116 GB118 GB119 GB123 GB
6-bit89.8 GB91.0 GB92.1 GB96.6 GB
5-bit77.9 GB79.1 GB80.2 GB84.7 GB
4-bit66.4 GB67.6 GB68.7 GB73.2 GB
3-bit53.5 GB54.7 GB55.8 GB60.3 GB
2-bit41.1 GB42.2 GB43.4 GB47.9 GB
1.6-bit22.6 GB23.8 GB24.9 GB29.4 GB

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

GPUs and Macs that run Laguna S 2.1

The best way Laguna S 2.1 runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: Laguna S 2.1 on RTX 4060 report
RTX 5060Too bigPDF: Laguna S 2.1 on RTX 5060 report
RTX 3070Too bigPDF: Laguna S 2.1 on RTX 3070 report
RTX 3080 10 GBToo bigPDF: Laguna S 2.1 on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: Laguna S 2.1 on RTX 3060 12 GB report
Arc B580Too bigPDF: Laguna S 2.1 on Arc B580 report
RTX 4070Too bigPDF: Laguna S 2.1 on RTX 4070 report
RTX 4070 SuperToo bigPDF: Laguna S 2.1 on RTX 4070 Super report
RTX 5070Too bigPDF: Laguna S 2.1 on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: Laguna S 2.1 on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: Laguna S 2.1 on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: Laguna S 2.1 on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: Laguna S 2.1 on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: Laguna S 2.1 on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: Laguna S 2.1 on RTX 4080 Super report
RTX 5070 TiToo bigPDF: Laguna S 2.1 on RTX 5070 Ti report
RTX 5080Too bigPDF: Laguna S 2.1 on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: Laguna S 2.1 on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: Laguna S 2.1 on RTX 3090 report
RTX 4090Too bigPDF: Laguna S 2.1 on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: Laguna S 2.1 on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: Laguna S 2.1 on RTX 5090 report
2× RTX 30902-bit64K130 tok/s0.9 sPDF: Laguna S 2.1 on 2× RTX 3090 report
2× RTX 40902-bit64K136 tok/s0.4 sPDF: Laguna S 2.1 on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB2-bit72K160 tok/s0.8 sPDF: Laguna S 2.1 on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada2-bit72K132 tok/s0.6 sPDF: Laguna S 2.1 on RTX 6000 Ada report
L40S2-bit72K123 tok/s1.0 sPDF: Laguna S 2.1 on L40S report
2× RTX 50903-bit120K162 tok/s0.3 sPDF: Laguna S 2.1 on 2× RTX 5090 report
4× RTX 30905-bit224K83 tok/s0.5 sPDF: Laguna S 2.1 on 4× RTX 3090 report
RTX PRO 6000 Blackwell5-bit256K130 tok/s0.4 sPDF: Laguna S 2.1 on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: Laguna S 2.1 on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: Laguna S 2.1 on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: Laguna S 2.1 on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: Laguna S 2.1 on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: Laguna S 2.1 on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB2-bit80K48 tok/s18 sPDF: Laguna S 2.1 on Mac mini M4 Pro 64 GB report
M4 Max 128 GB6-bit16K42 tok/s9.2 sPDF: Laguna S 2.1 on M4 Max 128 GB report
M5 Max 128 GB6-bit16K46 tok/s2.6 sPDF: Laguna S 2.1 on M5 Max 128 GB report
M3 Ultra 96 GB4-bit28K61 tok/s5.1 sPDF: Laguna S 2.1 on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit1M42 tok/s5.1 sPDF: Laguna S 2.1 on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit1M42 tok/s5.1 sPDF: Laguna S 2.1 on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit1M55 tok/s1.6 sPDF: Laguna S 2.1 on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB2-bit80K40 tok/s5.3 sPDF: Laguna S 2.1 on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB6-bit16K20 tok/s5.3 sPDF: Laguna S 2.1 on Ryzen AI Max+ 395 128 GB report
RTX Spark6-bit480K28 tok/s1.8 sPDF: Laguna S 2.1 on RTX Spark report
DGX Spark6-bit480K26 tok/s2.0 sPDF: Laguna S 2.1 on DGX Spark report
Dual DGX Spark8-bit1M20 tok/s1.5 sPDF: Laguna S 2.1 on Dual DGX Spark report
DGX Station GB3008-bit1M213 tok/s0.1 sPDF: Laguna S 2.1 on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit1M100 tok/s0.3 sPDF: Laguna S 2.1 on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit1M100 tok/s0.1 sPDF: Laguna S 2.1 on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit1M100 tok/s0.1 sPDF: Laguna S 2.1 on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM4-bit184K155 tok/s1.2 sPDF: Laguna S 2.1 on A100 80 GB SXM report
H100 SXM4-bit184K197 tok/s0.4 sPDF: Laguna S 2.1 on H100 SXM report
4× H200 NVL8-bit1M180 tok/s0.1 sPDF: Laguna S 2.1 on 4× H200 NVL report
8× H200 NVL8-bit1M180 tok/s0.1 sPDF: Laguna S 2.1 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 poolside/Laguna-S-2.1-GGUF
vLLM
vllm serve poolside/Laguna-S-2.1
MLX (Mac)
mlx_lm.chat --model mlx-community/Laguna-S-2.1-oQ4e

Questions

How much VRAM does Laguna S 2.1 need?

Laguna S 2.1 needs 66.4 GB for its weights at 4-bit, 67.6 GB with an 8K context and 68.7 GB with 32K. At 8-bit, 118 GB.

What is the minimum VRAM for Laguna S 2.1?

About 42.2 GB at 2-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 2× RTX 3090.

How big is Laguna S 2.1?

118B parameters (8.4B active per token): 219 GB of weights at 16-bit, 66.4 GB at 4-bit.

Can Laguna S 2.1 run on an RTX 4090?

Not on one card. The smallest machine here that runs it is the 2× RTX 3090.

More models