Skip to content

Molmo2 O 7B VRAM requirements

How much VRAM Molmo2 O 7B needs
4-bit weights: 4.4 GBWith 8K context: 9.1 GBWith 32K context: 21.1 GB8-bit, 8K context: 12.4 GB
Parameters
7.8B
Longest context
64K
Released
Dec 2025
License
Apache 2.0
Strongest at
Text and images

How much VRAM Molmo2 O 7B needs

Memory Molmo2 O 7B needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 64K
16-bit14.5 GB19.2 GB31.2 GB47.2 GB
8-bit7.7 GB12.4 GB24.4 GB40.4 GB
6-bit5.9 GB10.7 GB22.7 GB38.7 GB
5-bit5.1 GB9.9 GB21.9 GB37.9 GB
4-bit4.4 GB9.1 GB21.1 GB37.1 GB
3-bit3.5 GB8.3 GB20.3 GB36.3 GB
2-bit2.7 GB7.5 GB19.5 GB35.5 GB
1.6-bit1.5 GB6.3 GB18.3 GB34.3 GB

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

GPUs and Macs that run Molmo2 O 7B

The best way Molmo2 O 7B runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: Molmo2 O 7B on RTX 4060 report
RTX 5060Too bigPDF: Molmo2 O 7B on RTX 5060 report
RTX 3070Too bigPDF: Molmo2 O 7B on RTX 3070 report
RTX 3080 10 GBToo bigPDF: Molmo2 O 7B on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: Molmo2 O 7B on RTX 3060 12 GB report
Arc B580Too bigPDF: Molmo2 O 7B on Arc B580 report
RTX 4070Too bigPDF: Molmo2 O 7B on RTX 4070 report
RTX 4070 SuperToo bigPDF: Molmo2 O 7B on RTX 4070 Super report
RTX 5070Too bigPDF: Molmo2 O 7B on RTX 5070 report
RTX 4060 Ti 16 GB6-bit16K32 tok/s0.7 sPDF: Molmo2 O 7B on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB6-bit16K50 tok/s0.7 sPDF: Molmo2 O 7B on RTX 5060 Ti 16 GB report
Arc A770 16 GB6-bit16K43 tok/s0.8 sPDF: Molmo2 O 7B on Arc A770 16 GB report
Radeon RX 9070 XT6-bit16K58 tok/s0.4 sPDF: Molmo2 O 7B on Radeon RX 9070 XT report
RTX 4070 Ti Super6-bit16K73 tok/s0.4 sPDF: Molmo2 O 7B on RTX 4070 Ti Super report
RTX 4080 Super6-bit16K79 tok/s0.3 sPDF: Molmo2 O 7B on RTX 4080 Super report
RTX 5070 Ti6-bit16K95 tok/s0.4 sPDF: Molmo2 O 7B on RTX 5070 Ti report
RTX 50806-bit16K101 tok/s0.3 sPDF: Molmo2 O 7B on RTX 5080 report
Radeon RX 7900 XTX8-bit28K66 tok/s0.7 sPDF: Molmo2 O 7B on Radeon RX 7900 XTX report
RTX 30908-bit28K78 tok/s0.4 sPDF: Molmo2 O 7B on RTX 3090 report
RTX 40908-bit28K83 tok/s0.2 sPDF: Molmo2 O 7B on RTX 4090 report
Radeon AI PRO R97008-bit40K46 tok/s0.4 sPDF: Molmo2 O 7B on Radeon AI PRO R9700 report
RTX 50908-bit40K139 tok/s0.1 sPDF: Molmo2 O 7B on RTX 5090 report
2× RTX 30908-bit64K78 tok/s0.3 sPDF: Molmo2 O 7B on 2× RTX 3090 report
2× RTX 40908-bit64K83 tok/s0.1 sPDF: Molmo2 O 7B on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit64K108 tok/s0.2 sPDF: Molmo2 O 7B on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit64K79 tok/s0.2 sPDF: Molmo2 O 7B on RTX 6000 Ada report
L40S8-bit64K72 tok/s0.3 sPDF: Molmo2 O 7B on L40S report
2× RTX 50908-bit64K139 tok/s0.1 sPDF: Molmo2 O 7B on 2× RTX 5090 report
4× RTX 30908-bit64K78 tok/s0.1 sPDF: Molmo2 O 7B on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit64K139 tok/s0.1 sPDF: Molmo2 O 7B on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: Molmo2 O 7B on Mac mini M4 16 GB report
Mac mini M4 24 GB6-bit16K16 tok/s9.9 sPDF: Molmo2 O 7B on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit24K12 tok/s9.9 sPDF: Molmo2 O 7B on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB6-bit16K34 tok/s4.9 sPDF: Molmo2 O 7B on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit48K27 tok/s4.9 sPDF: Molmo2 O 7B on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit64K27 tok/s4.9 sPDF: Molmo2 O 7B on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit64K46 tok/s2.5 sPDF: Molmo2 O 7B on M4 Max 128 GB report
M5 Max 128 GB8-bit64K51 tok/s0.7 sPDF: Molmo2 O 7B on M5 Max 128 GB report
M3 Ultra 96 GB8-bit64K58 tok/s1.4 sPDF: Molmo2 O 7B on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit64K58 tok/s1.4 sPDF: Molmo2 O 7B on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit64K58 tok/s1.4 sPDF: Molmo2 O 7B on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit64K80 tok/s0.4 sPDF: Molmo2 O 7B on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit64K19 tok/s1.4 sPDF: Molmo2 O 7B on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit64K19 tok/s1.4 sPDF: Molmo2 O 7B on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit64K26 tok/s0.5 sPDF: Molmo2 O 7B on RTX Spark report
DGX Spark8-bit64K24 tok/s0.5 sPDF: Molmo2 O 7B on DGX Spark report
Dual DGX Spark8-bit64K24 tok/s0.3 sPDF: Molmo2 O 7B on Dual DGX Spark report
DGX Station GB3008-bit64K395 tok/s0.1 sPDF: Molmo2 O 7B on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit64K139 tok/s0.1 sPDF: Molmo2 O 7B on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit64K139 tok/s0.1 sPDF: Molmo2 O 7B on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit64K139 tok/s0.1 sPDF: Molmo2 O 7B on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit64K155 tok/s0.3 sPDF: Molmo2 O 7B on A100 80 GB SXM report
H100 SXM8-bit64K233 tok/s0.1 sPDF: Molmo2 O 7B on H100 SXM report
4× H200 NVL8-bit64K304 tok/s0.1 sPDF: Molmo2 O 7B on 4× H200 NVL report
8× H200 NVL8-bit64K304 tok/s0.1 sPDF: Molmo2 O 7B 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 mu2solutions/Molmo2-O-7B-GGUF
vLLM
vllm serve allenai/Molmo2-O-7B

Questions

How much VRAM does Molmo2 O 7B need?

Molmo2 O 7B needs 4.4 GB for its weights at 4-bit, 9.1 GB with an 8K context and 21.1 GB with 32K. At 8-bit, 12.4 GB.

What is the minimum VRAM for Molmo2 O 7B?

About 8.3 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 Ti 16 GB.

How big is Molmo2 O 7B?

7.8B parameters: 14.5 GB of weights at 16-bit, 4.4 GB at 4-bit.

Can Molmo2 O 7B run on an RTX 4090?

Yes. At 8-bit it holds 28K of context and writes about 83 tokens a second.

More models