How much VRAM does Devstral 2 123B Instruct need?

Devstral 2 123B Instruct (123B parameters) requires approximately 110.5 GB of memory with Q4_K_M quantization.

What is the best quantization for Devstral 2 123B Instruct?

The recommended quantization for Devstral 2 123B Instruct is Q4_K_M, which balances quality and memory efficiency.

Can it run?

Yes, MacBook Pro M3 Max 128GB can run Devstral 2 123B Instruct with a D grade (Very compromised). Expected decode speed: 3.3 tok/s.

DPoor

Very compromised

Using Q4_K_M in vLLM

Capabilities:

Fit status

Very compromised

Decode

3.3 tok/s

TTFT

58792 ms

Safe context

13K

Memory

110.5 GB / 92.2 GB

Offload

20%

Weights75.0 GB

KV Cache19.2 GB

Runtime2.4 GB

Headroom13.8 GB

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	F	Too heavy	3.2 tok/s	88035 ms	23K
Chat	D	Very compromised (needs ~6.5 GB host RAM)	3.1 tok/s	33891 ms	7K
Coding	D	Very compromised	3.3 tok/s	58792 ms	13K
RAG	F	Too heavy	3.2 tok/s	110044 ms	23K
Reasoning	D	Very compromised (needs ~12.4 GB host RAM)	3.0 tok/s	75272 ms

How Devstral 2 123B Instruct (123B params) fits at each quantization level on MacBook Pro M3 Max 128GB (92.2 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	48.0 GB	Low	C41
Q3_K_S	3	60.3 GB	Low	C44
NVFP4Best for your GPU

HuggingFace

huggingface-cli download devstral-2-123b

Upgrade options

AMD Instinct MI350X 288GBBudget pick

C77.8 tok/s decode

~$8,000 MSRP

AMD Instinct MI300X 192GBBest value

C55.1 tok/s decode

~$15,000 MSRP

NVIDIA B200 180GBBiggest leap