Can Mistral Small 4 119B run on NVIDIA L40 48GB?

Q: What should I upgrade first if Mistral Small 4 119B feels slow on NVIDIA L40 48GB?

Remove offload with more accelerator memory. Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.

YES — With Q2_K

A73Great

Estimated — low-sample bucket· few comparable runs

Mistral Small 4 119B needs ~57.5 GB VRAM. NVIDIA L40 48GB has 48.0 GB. With Q2_K quantization, expect ~20 tok/s.

Runtime: llama.cppCapacity: OffloadBandwidth: HighStack: StandardBottleneck: Host offload

Operating mode

Choose the run profile you care about

Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.

Current mode

Balanced

Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.

Mistral Small 4 119B at Q4_K_M needs 83.7 GB — too much for NVIDIA L40 48GB (48.0 GB). Runs at Q2_K (57.5 GB) with low quality.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 83.7 GB, exceeds 48.0 GB available

83.7 GB required48.0 GB available

174% VRAM needed

35.7 GB over capacity — needs offload or smaller quantization

Fit status

Too heavy

Decode

6.8 tok/s

TTFT

28298 ms

Safe context

Memory

83.7 GB / 48.0 GB

Offload

40%

Memory breakdown

Weights72.6 GB

KV Cache5.4 GB

Runtime0.9 GB

Headroom4.8 GB

See how fast it feels

With memory offload — actual speed may be lower

See how fast it feelsMistral Small 4 119B on NVIDIA L40 48GB
1st promptCold start — includes initialization
>Write a Python function to sort a list
Here's a Python function that sorts a list using quicksort:

def quicksort(arr):
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

# Example usage
numbers = [3, 6, 8, 10, 1, 2, 1]
print(quicksort(numbers))  # [1, 1, 2, 3, 6, 8, 10]

The algorithm picks a pivot, partitions the list into smaller, equal, and larger elements, then recursively sorts each partition and concatenates the results.
2nd promptWarm — model ready, no init overhead
>Now add type hints and a docstring
def quicksort(arr: list[int]) -> list[int]:
    """Sort a list using the quicksort algorithm.

    Args:
        arr: List of integers to sort.

    Returns:
        A new sorted list.
    """
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

Added type hints for the input and return type, plus a Google-style docstring.
Estimated: 6.8 tok/s decode · 28.3s TTFT (warm) · 17 tok/s prefill

What limits this setup

It fits through host-memory offload, and offload is the main reason performance drops.

CPU or host-memory offload is active

About 20% of the working set spills out of accelerator memory, which usually hurts latency and sustained decode throughput.

Very little memory headroom

You can run the model, but there is not much room left for longer context, bigger batches, extra apps, or future model updates.

Best improvement path

Remove offload with more accelerator memory

Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.

Buy headroom, not only minimum fit

A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.

Increase host RAM if you keep offloading

This setup may need roughly 7.7 GB of extra host RAM just for the offloaded portion, before OS and other tools.

Performance by workload

Workload	Grade	Fit	Decode	TTFT	Context
Chat	F	Too heavy	7.3 tok/s	14411 ms	4K
Coding	F	Too heavy	6.8 tok/s	28298 ms	4K
Agentic Coding	F	Too heavy	6.0 tok/s	46921 ms	4K
Reasoning	F	Too heavy	6.8 tok/s	33443 ms	4K
RAG	F	Too heavy	6.0 tok/s	58652 ms	4K

Quantization options

How Mistral Small 4 119B (119B params) fits at each quantization level on NVIDIA L40 48GB (48.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	46.4 GB	Low	F0
Q3_K_S	3	58.3 GB	Low	F0
NVFP4	4

Get started

Copy-paste commands to run Mistral Small 4 119B on your machine.

Run

lms load Mistral-Small-4-119B-2603 && lms server start

Upgrade options

Hardware that runs Mistral Small 4 119B well

RTX PRO 6000 Blackwell Workstation Edition 96GBBudget pick

96 GB VRAM (+48)1792 GB/s (+928)

Makes the model fit on the accelerator instead of staying completely out of reach.65.6 tok/s decode

Makes the model fit on the accelerator instead of staying completely out of reach.

Removes host-memory offload, which is usually the single biggest latency and throughput win.

~$9,999 MSRP

RTX PRO 6000 Blackwell Server Edition 96GBBest value

96 GB VRAM (+48)1597 GB/s (+733)

Makes the model fit on the accelerator instead of staying completely out of reach.58.5 tok/s decode

Makes the model fit on the accelerator instead of staying completely out of reach.

Removes host-memory offload, which is usually the single biggest latency and throughput win.

~$9,999 MSRP

NVIDIA H20 96GBNVIDIA upgrade

96 GB VRAM (+48)4000 GB/s (+3136)

Makes the model fit on the accelerator instead of staying completely out of reach.141.2 tok/s decode

Makes the model fit on the accelerator instead of staying completely out of reach.

Removes host-memory offload, which is usually the single biggest latency and throughput win.

~$12,000 MSRP

Frequently asked questions

Can NVIDIA L40 48GB run Mistral Small 4 119B?

Yes, NVIDIA L40 48GB can run Mistral Small 4 119B at Q2_K quantization (Very compromised (needs ~7.7 GB host RAM)). The recommended Q4_K_M requires 83.7 GB which exceeds available memory, but at Q2_K it needs only 57.5 GB. Expected decode speed: 20.0 tok/s.

How much VRAM does Mistral Small 4 119B need?

Mistral Small 4 119B (119B parameters) requires approximately 83.7 GB at Q4_K_M quantization. On NVIDIA L40 48GB, it fits at Q2_K using 57.5 GB.

What is the best quantization for Mistral Small 4 119B?

The recommended quantization is Q4_K_M, but on NVIDIA L40 48GB the best fitting quantization is Q2_K, which uses 57.5 GB.

What speed will Mistral Small 4 119B run at on NVIDIA L40 48GB?

On NVIDIA L40 48GB, Mistral Small 4 119B achieves approximately 20.0 tokens per second decode speed with a time-to-first-token of 9660ms using Q2_K quantization.

Can NVIDIA L40 48GB run Mistral Small 4 119B for coding?

For coding workloads, Mistral Small 4 119B on NVIDIA L40 48GB receives a F grade with 6.8 tok/s and 4K context.

What context window can Mistral Small 4 119B use on NVIDIA L40 48GB?

On NVIDIA L40 48GB, Mistral Small 4 119B can safely use up to 4K tokens of context at Q2_K quantization. The model's official context limit is 256K, but available memory constrains the safe maximum.

See all results for NVIDIA L40 48GB See all hardware for Mistral Small 4 119B

Embed this result▼

Paste this snippet into any page to show a live fit card.

<iframe src="https://willitrunai.com/embed/mistral-small-4-119b-on-l40-48gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview: