Can Codestral 22B v0.1 run on Mac Studio M3 Ultra 96GB?

YES — Runs Great

C48Usable

○Estimated from fit model

Codestral 22B v0.1 needs ~27.3 GB VRAM. Mac Studio M3 Ultra 96GB has 69.1 GB. With Q4_K_M quantization, expect ~42 tok/s.

Runtime: llama.cppCapacity: RoomyBandwidth: HighStack: StandardBottleneck: Balanced

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.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 27.3 GB, 41.5 tok/s, Runs well

27.3 GB required69.1 GB available

40% VRAM used

Fit status

Runs well

Decode

41.5 tok/s

TTFT

4665 ms

Safe context

276K

Memory

27.3 GB / 69.1 GB

Memory breakdown

Weights13.4 GB

KV Cache2.6 GB

Runtime0.9 GB

Headroom10.4 GB

See how fast it feels

See how fast it feelsCodestral 22B v0.1 on Mac Studio M3 Ultra 96GB
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: 41.5 tok/s decode · 4.7s TTFT (warm) · 104 tok/s prefill

What limits this setup

This setup is broadly balanced for this model.

Shared-memory contention still exists

The OS, browser, and inference runtime all compete for the same physical memory pool, so real-world headroom is less forgiving than raw capacity suggests.

Best improvement path

Performance by workload

Workload	Grade	Fit	Decode	TTFT	Context
Chat	C	Runs well	41.5 tok/s	2545 ms	276K
Coding	C	Runs well	41.5 tok/s	4665 ms	276K
Agentic Coding	C	Runs well	41.5 tok/s	6786 ms	276K
Reasoning	C	Runs well	41.5 tok/s	5513 ms	276K
RAG	C	Runs well	41.5 tok/s	8482 ms	276K

Quantization options

How Codestral 22B v0.1 (22B params) fits at each quantization level on Mac Studio M3 Ultra 96GB (69.1 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	8.6 GB	Low	C41
Q3_K_S	3	10.8 GB	Low	C41
NVFP4	4	12.3 GB	Medium	C41
Q4_K_M	4	13.4 GB	Medium	C41
Q5_K_M	5	15.8 GB	High	C42
Q6_K	6	18.0 GB	High	C42
Q8_0	8	23.5 GB	Very High	C43
F16Best for your GPU	16	45.1 GB	Maximum	C48

Get started

Copy-paste commands to run Codestral 22B v0.1 on your machine.

Run

lms load hf-bartowski--codestral-22b-v0-1-gguf && lms server start

Upgrade options

Hardware that runs Codestral 22B v0.1 well

👁 NVIDIA
RTX PRO 6000 Blackwell Workstation Edition 96GBBudget pick

1792 GB/s (+973)

Raises estimated decode speed by about 170%.112.2 tok/s decode

Raises estimated decode speed by about 170%.

Adds memory headroom for longer context windows and future model growth.

~$9,999 MSRP

👁 NVIDIA
RTX PRO 6000 Blackwell Server Edition 96GBBest value

1597 GB/s (+778)

Raises estimated decode speed by about 141%.100 tok/s decode

Raises estimated decode speed by about 141%.