Skip to content

MiniMax M3 VRAM requirements

How much VRAM MiniMax M3 needs
4-bit weights: 241 GBWith 8K context: 243 GBWith 32K context: 246 GB8-bit, 8K context: 425 GB
Parameters
427B
Active per token
23B
Longest context
1M
Released
Jun 2026
License
MiniMax Community
Strongest at
Docs

How much VRAM MiniMax M3 needs

Memory MiniMax M3 needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit795 GB797 GB800 GB811 GB
8-bit423 GB425 GB427 GB439 GB
6-bit326 GB328 GB331 GB342 GB
5-bit283 GB285 GB288 GB299 GB
4-bit241 GB243 GB246 GB257 GB
3-bit194 GB196 GB199 GB210 GB
2-bit149 GB151 GB154 GB165 GB
1.6-bit82.0 GB84.0 GB86.8 GB98.0 GB

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

GPUs and Macs that run MiniMax M3

The best way MiniMax M3 runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: MiniMax M3 on RTX 4060 report
RTX 5060Too bigPDF: MiniMax M3 on RTX 5060 report
RTX 3070Too bigPDF: MiniMax M3 on RTX 3070 report
RTX 3080 10 GBToo bigPDF: MiniMax M3 on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: MiniMax M3 on RTX 3060 12 GB report
Arc B580Too bigPDF: MiniMax M3 on Arc B580 report
RTX 4070Too bigPDF: MiniMax M3 on RTX 4070 report
RTX 4070 SuperToo bigPDF: MiniMax M3 on RTX 4070 Super report
RTX 5070Too bigPDF: MiniMax M3 on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: MiniMax M3 on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: MiniMax M3 on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: MiniMax M3 on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: MiniMax M3 on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: MiniMax M3 on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: MiniMax M3 on RTX 4080 Super report
RTX 5070 TiToo bigPDF: MiniMax M3 on RTX 5070 Ti report
RTX 5080Too bigPDF: MiniMax M3 on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: MiniMax M3 on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: MiniMax M3 on RTX 3090 report
RTX 4090Too bigPDF: MiniMax M3 on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: MiniMax M3 on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: MiniMax M3 on RTX 5090 report
2× RTX 3090Too bigPDF: MiniMax M3 on 2× RTX 3090 report
2× RTX 4090Too bigPDF: MiniMax M3 on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GBToo bigPDF: MiniMax M3 on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 AdaToo bigPDF: MiniMax M3 on RTX 6000 Ada report
L40SToo bigPDF: MiniMax M3 on L40S report
2× RTX 5090Too bigPDF: MiniMax M3 on 2× RTX 5090 report
4× RTX 30901.6-bit48K92 tok/s1.3 sPDF: MiniMax M3 on 4× RTX 3090 report
RTX PRO 6000 Blackwell1.6-bit64K135 tok/s1.2 sPDF: MiniMax M3 on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: MiniMax M3 on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: MiniMax M3 on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: MiniMax M3 on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: MiniMax M3 on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: MiniMax M3 on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GBToo bigPDF: MiniMax M3 on Mac mini M4 Pro 64 GB report
M4 Max 128 GB1.6-bit64K50 tok/s24 sPDF: MiniMax M3 on M4 Max 128 GB report
M5 Max 128 GB1.6-bit64K53 tok/s7.0 sPDF: MiniMax M3 on M5 Max 128 GB report
M3 Ultra 96 GBToo bigPDF: MiniMax M3 on M3 Ultra 96 GB report
M3 Ultra 256 GB2-bit256K41 tok/s14 sPDF: MiniMax M3 on M3 Ultra 256 GB report
M3 Ultra 512 GB6-bit320K23 tok/s14 sPDF: MiniMax M3 on M3 Ultra 512 GB report
M5 Ultra 512 GB6-bit320K31 tok/s4.3 sPDF: MiniMax M3 on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GBToo bigPDF: MiniMax M3 on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB1.6-bit64K28 tok/s14 sPDF: MiniMax M3 on Ryzen AI Max+ 395 128 GB report
RTX Spark1.6-bit256K38 tok/s4.9 sPDF: MiniMax M3 on RTX Spark report
DGX Spark1.6-bit256K35 tok/s4.9 sPDF: MiniMax M3 on DGX Spark report
Dual DGX Spark3-bit256K16 tok/s2.8 sPDF: MiniMax M3 on Dual DGX Spark report
DGX Station GB3003-bit352K171 tok/s0.4 sPDF: MiniMax M3 on DGX Station GB300 report
2× RTX PRO 6000 Blackwell2-bit256K95 tok/s0.7 sPDF: MiniMax M3 on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell6-bit288K54 tok/s0.4 sPDF: MiniMax M3 on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit1M43 tok/s0.2 sPDF: MiniMax M3 on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXMToo bigPDF: MiniMax M3 on A100 80 GB SXM report
H100 SXMToo bigPDF: MiniMax M3 on H100 SXM report
4× H200 NVL8-bit928K92 tok/s0.4 sPDF: MiniMax M3 on 4× H200 NVL report
8× H200 NVL8-bit1M92 tok/s0.2 sPDF: MiniMax M3 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 unsloth/MiniMax-M3-GGUF
vLLM
vllm serve MiniMaxAI/MiniMax-M3
MLX (Mac)
mlx_lm.chat --model mlx-community/MiniMax-M3-4bit

Questions

How much VRAM does MiniMax M3 need?

MiniMax M3 needs 241 GB for its weights at 4-bit, 243 GB with an 8K context and 246 GB with 32K. At 8-bit, 425 GB.

What is the minimum VRAM for MiniMax M3?

About 84.0 GB at 1.6-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 4× RTX 3090.

How big is MiniMax M3?

427B parameters (23B active per token): 795 GB of weights at 16-bit, 241 GB at 4-bit.

Can MiniMax M3 run on an RTX 4090?

Not on one card. The smallest machine here that runs it is the 4× RTX 3090.

More models