Skip to content

MiMo 7B RL VRAM requirements

How much VRAM MiMo 7B RL needs
4-bit weights: 4.4 GBWith 8K context: 6.4 GBWith 32K context: 9.8 GB8-bit, 8K context: 9.7 GB
Parameters
7.8B
Longest context
32K
Released
Apr 2025
License
MIT
Strongest at
Not rated

How much VRAM MiMo 7B RL needs

Memory MiMo 7B RL needs at each precision and context length
PrecisionWeightsWith 8KWith 32K
16-bit14.6 GB16.6 GB20.0 GB
8-bit7.8 GB9.7 GB13.1 GB
6-bit6.0 GB8.0 GB11.3 GB
5-bit5.2 GB7.2 GB10.6 GB
4-bit4.4 GB6.4 GB9.8 GB
3-bit3.6 GB5.6 GB8.9 GB
2-bit2.7 GB4.7 GB8.1 GB
1.6-bit1.5 GB3.5 GB6.9 GB

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

GPUs and Macs that run MiMo 7B RL

The best way MiMo 7B RL runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40604-bit16K42 tok/s1.0 sPDF: MiMo 7B RL on RTX 4060 report
RTX 50604-bit16K67 tok/s0.8 sPDF: MiMo 7B RL on RTX 5060 report
RTX 30704-bit16K67 tok/s0.7 sPDF: MiMo 7B RL on RTX 3070 report
RTX 3080 10 GB6-bit16K82 tok/s0.5 sPDF: MiMo 7B RL on RTX 3080 10 GB report
RTX 3060 12 GB8-bit16K32 tok/s1.2 sPDF: MiMo 7B RL on RTX 3060 12 GB report
Arc B5808-bit16K28 tok/s0.8 sPDF: MiMo 7B RL on Arc B580 report
RTX 40708-bit16K44 tok/s0.5 sPDF: MiMo 7B RL on RTX 4070 report
RTX 4070 Super8-bit16K44 tok/s0.4 sPDF: MiMo 7B RL on RTX 4070 Super report
RTX 50708-bit16K58 tok/s0.5 sPDF: MiMo 7B RL on RTX 5070 report
RTX 4060 Ti 16 GB8-bit32K26 tok/s0.7 sPDF: MiMo 7B RL on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit32K39 tok/s0.6 sPDF: MiMo 7B RL on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit32K34 tok/s0.7 sPDF: MiMo 7B RL on Arc A770 16 GB report
Radeon RX 9070 XT8-bit32K46 tok/s0.4 sPDF: MiMo 7B RL on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit32K58 tok/s0.3 sPDF: MiMo 7B RL on RTX 4070 Ti Super report
RTX 4080 Super8-bit32K63 tok/s0.3 sPDF: MiMo 7B RL on RTX 4080 Super report
RTX 5070 Ti8-bit32K75 tok/s0.3 sPDF: MiMo 7B RL on RTX 5070 Ti report
RTX 50808-bit32K80 tok/s0.3 sPDF: MiMo 7B RL on RTX 5080 report
Radeon RX 7900 XTX8-bit32K66 tok/s0.7 sPDF: MiMo 7B RL on Radeon RX 7900 XTX report
RTX 30908-bit32K78 tok/s0.4 sPDF: MiMo 7B RL on RTX 3090 report
RTX 40908-bit32K84 tok/s0.2 sPDF: MiMo 7B RL on RTX 4090 report
Radeon AI PRO R97008-bit32K46 tok/s0.4 sPDF: MiMo 7B RL on Radeon AI PRO R9700 report
RTX 50908-bit32K139 tok/s0.1 sPDF: MiMo 7B RL on RTX 5090 report
2× RTX 30908-bit32K78 tok/s0.2 sPDF: MiMo 7B RL on 2× RTX 3090 report
2× RTX 40908-bit32K84 tok/s0.1 sPDF: MiMo 7B RL on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit32K108 tok/s0.2 sPDF: MiMo 7B RL on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit32K80 tok/s0.2 sPDF: MiMo 7B RL on RTX 6000 Ada report
L40S8-bit32K73 tok/s0.3 sPDF: MiMo 7B RL on L40S report
2× RTX 50908-bit32K139 tok/s0.1 sPDF: MiMo 7B RL on 2× RTX 5090 report
4× RTX 30908-bit32K78 tok/s0.1 sPDF: MiMo 7B RL on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit32K139 tok/s0.1 sPDF: MiMo 7B RL on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB6-bit24K16 tok/s9.4 sPDF: MiMo 7B RL on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit32K12 tok/s9.4 sPDF: MiMo 7B RL on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit32K12 tok/s9.4 sPDF: MiMo 7B RL on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit32K27 tok/s4.7 sPDF: MiMo 7B RL on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit32K27 tok/s4.7 sPDF: MiMo 7B RL on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit32K27 tok/s4.7 sPDF: MiMo 7B RL on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit32K46 tok/s2.4 sPDF: MiMo 7B RL on M4 Max 128 GB report
M5 Max 128 GB8-bit32K51 tok/s0.7 sPDF: MiMo 7B RL on M5 Max 128 GB report
M3 Ultra 96 GB8-bit32K58 tok/s1.3 sPDF: MiMo 7B RL on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit32K58 tok/s1.3 sPDF: MiMo 7B RL on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit32K58 tok/s1.3 sPDF: MiMo 7B RL on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit32K79 tok/s0.4 sPDF: MiMo 7B RL on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit32K19 tok/s1.4 sPDF: MiMo 7B RL on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit32K19 tok/s1.4 sPDF: MiMo 7B RL on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit32K27 tok/s0.5 sPDF: MiMo 7B RL on RTX Spark report
DGX Spark8-bit32K24 tok/s0.5 sPDF: MiMo 7B RL on DGX Spark report
Dual DGX Spark8-bit32K24 tok/s0.3 sPDF: MiMo 7B RL on Dual DGX Spark report
DGX Station GB3008-bit32K381 tok/s0.1 sPDF: MiMo 7B RL on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit32K139 tok/s0.1 sPDF: MiMo 7B RL on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit32K139 tok/s0.1 sPDF: MiMo 7B RL on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit32K139 tok/s0.1 sPDF: MiMo 7B RL on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit32K155 tok/s0.3 sPDF: MiMo 7B RL on A100 80 GB SXM report
H100 SXM8-bit32K230 tok/s0.1 sPDF: MiMo 7B RL on H100 SXM report
4× H200 NVL8-bit32K298 tok/s0.1 sPDF: MiMo 7B RL on 4× H200 NVL report
8× H200 NVL8-bit32K298 tok/s0.1 sPDF: MiMo 7B RL 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 jedisct1/MiMo-7B-RL-GGUF
vLLM
vllm serve XiaomiMiMo/MiMo-7B-RL
MLX (Mac)
mlx_lm.chat --model mlx-community/MiMo-7B-RL-4bit

Questions

How much VRAM does MiMo 7B RL need?

MiMo 7B RL needs 4.4 GB for its weights at 4-bit, 6.4 GB with an 8K context and 9.8 GB with 32K. At 8-bit, 9.7 GB.

What is the minimum VRAM for MiMo 7B RL?

About 5.6 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 MiMo 7B RL?

7.8B parameters: 14.6 GB of weights at 16-bit, 4.4 GB at 4-bit.

Can MiMo 7B RL run on an RTX 4090?

Yes. At 8-bit it holds 32K of context and writes about 84 tokens a second.

More models