How much VRAM does Devstral 2 123B Instruct need?

Devstral 2 123B Instruct (123B parameters) requires approximately 109.4 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, AMD Instinct MI250X 128GB can run Devstral 2 123B Instruct with a C grade (Tight fit). Expected decode speed: 33.3 tok/s.

CUsable

Tight fit

Using Q4_K_M in vLLM

Capabilities:

Fit status

Tight fit

Decode

33.3 tok/s

TTFT

5820 ms

Safe context

19K

Memory

109.4 GB / 128.0 GB

Weights75.0 GB

KV Cache19.2 GB

Runtime2.4 GB

Headroom12.8 GB

Workload	Grade	Fit	Decode	TTFT	Context
Agentic Coding	C	Runs with offload (needs ~0.4 GB host RAM)	33.1 tok/s	8502 ms	32K
Chat	C	Runs well	33.3 tok/s	3174 ms	10K
Coding	C	Tight fit	33.3 tok/s	5820 ms	19K
RAG	C	Runs with offload (needs ~0.4 GB host RAM)	33.1 tok/s	10628 ms	32K
Reasoning	C	Tight fit	33.3 tok/s	6878 ms

How Devstral 2 123B Instruct (123B params) fits at each quantization level on AMD Instinct MI250X 128GB (128.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	48.0 GB	Low	D38
Q3_K_S	3	60.3 GB	Low	D40
NVFP4	4

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