Can Codestral 21B Pruned i1 run on MacBook Pro M4 Pro 48GB?

YES — Runs Great

C50Usable

Estimated — low-sample bucket· few comparable runs

Codestral 21B Pruned i1 needs ~21.4 GB VRAM. MacBook Pro M4 Pro 48GB has 34.6 GB. With Q4_K_M quantization, expect ~22 tok/s.

Runtime: llama.cppCapacity: RoomyBandwidth: LowStack: 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) — 21.4 GB, 22.2 tok/s, Runs well

21.4 GB required34.6 GB available

62% VRAM used

Fit status

Runs well

Decode

22.2 tok/s

TTFT

8739 ms

Safe context

102K

Memory

21.4 GB / 34.6 GB

Memory breakdown

Weights12.8 GB

KV Cache2.5 GB

Runtime0.9 GB

Headroom5.2 GB

See how fast it feels

See how fast it feelsCodestral 21B Pruned i1 on MacBook Pro M4 Pro 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: 22.2 tok/s decode · 8.7s TTFT (warm) · 55 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	22.2 tok/s	4767 ms	102K
Coding	C	Runs well	22.2 tok/s	8739 ms	102K
Agentic Coding	C	Runs well	22.2 tok/s	12711 ms	102K
Reasoning	C	Runs well	22.2 tok/s	10328 ms	102K
RAG	C	Runs well	22.2 tok/s	15889 ms	102K

Inference speed

Codestral 21B Pruned i1 inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for Codestral 21B Pruned i1 at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~94 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.

GPU / Mac	Memory	Quant	Speed (tok/s)	Fits?
RTX 5090 32GB	32 GB	Q4_K_M	93.7	Fits
RTX 4090 24GB	24 GB	Q4_K_M	59.8	Fits
RX 7900 XTX 24GB	24 GB	Q4_K_M	54.0	Fits
RTX 3090 24GB	24 GB	Q4_K_M	51.1	Fits
Mac Studio M3 Ultra 256GB	256 GB	Q4_K_M	43.5	Fits
Mac Studio M2 Ultra 128GB	128 GB	Q4_K_M	36.2	Fits
MacBook Pro M4 Max 128GB	128 GB	Q4_K_M	35.2	Fits
MacBook Pro M4 Max 64GB	64 GB	Q4_K_M	35.2	Fits
Mac Studio M1 Ultra 128GB	128 GB	Q4_K_M	34.3	Fits
RTX 4080 Super 16GB	16 GB	Q4_K_M	27.7	Heavy offload
MacBook Pro M4 Pro 48GB	48 GB	Q4_K_M	22.2	Fits
MacBook Pro M3 Max 64GB	64 GB	Q4_K_M	18.7	Fits
MacBook Pro M1 Max 64GB	64 GB	Q4_K_M	17.2	Fits
RTX 4070 12GB	12 GB	Q4_K_M	9.8	Too big
RTX 3060 12GB	12 GB	Q4_K_M	6.2	Too big
RTX 4060 8GB	8 GB	Q4_K_M	2.3	Too big

Estimates for single-stream decoding at Q4_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.

Quantization options

How Codestral 21B Pruned i1 (21B params) fits at each quantization level on MacBook Pro M4 Pro 48GB (34.6 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	8.2 GB	Low	C44
Q3_K_S	3	10.3 GB	Low	C45
NVFP4	4	11.8 GB	Medium	C46
Q4_K_M	4	12.8 GB	Medium	C46
Q5_K_M	5	15.1 GB	High	C47
Q6_K	6	17.2 GB	High	C48
Q8_0Best for your GPU	8	22.5 GB	Very High	C48
F16	16	43.1 GB	Maximum	F0

Get started

Copy-paste commands to run Codestral 21B Pruned i1 on your machine.

Run

lms load hf-mradermacher--codestral-21b-pruned-i1-gguf && lms server start

Opções de upgrade

Hardware que roda bem Codestral 21B Pruned i1

RTX PRO 5000 Blackwell 48GBOpção econômica

1344 GB/s (+1071)

Raises estimated decode speed by about 297%.88.1 tok/s decodificação

Raises estimated decode speed by about 297%.

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

~$4,999 MSRP

NVIDIA A100 40GBMelhor custo-benefício

1555 GB/s (+1282)

Raises estimated decode speed by about 359%.102 tok/s decodificação

Raises estimated decode speed by about 359%.

~$10,000 MSRP

Frequently asked questions

Can MacBook Pro M4 Pro 48GB run Codestral 21B Pruned i1?

Yes, MacBook Pro M4 Pro 48GB can run Codestral 21B Pruned i1 with a C grade (Runs well). Expected decode speed: 22.2 tok/s.

How much VRAM does Codestral 21B Pruned i1 need?

Codestral 21B Pruned i1 (21B parameters) requires approximately 21.4 GB of memory with Q4_K_M quantization.

What is the best quantization for Codestral 21B Pruned i1?

The recommended quantization for Codestral 21B Pruned i1 is Q4_K_M, which balances quality and memory efficiency.

What speed will Codestral 21B Pruned i1 run at on MacBook Pro M4 Pro 48GB?

On MacBook Pro M4 Pro 48GB, Codestral 21B Pruned i1 achieves approximately 22.2 tokens per second decode speed with a time-to-first-token of 8739ms using Q4_K_M quantization.

Can MacBook Pro M4 Pro 48GB run Codestral 21B Pruned i1 for coding?

For coding workloads, Codestral 21B Pruned i1 on MacBook Pro M4 Pro 48GB receives a C grade with 22.2 tok/s and 102K context.

What context window can Codestral 21B Pruned i1 use on MacBook Pro M4 Pro 48GB?

On MacBook Pro M4 Pro 48GB, Codestral 21B Pruned i1 can safely use up to 102K tokens of context. The model's official context limit is —, but available memory constrains the safe maximum.

Is unified memory on MacBook Pro M4 Pro 48GB as fast as VRAM for Codestral 21B Pruned i1?

Not always. MacBook Pro M4 Pro 48GB can often fit larger models thanks to unified memory, but a discrete GPU with dedicated high-bandwidth VRAM may still decode faster once the model fits. For this combination, the important distinction is capacity versus sustained throughput.

See all results for MacBook Pro M4 Pro 48GB See all hardware for Codestral 21B Pruned i1

Embed this result▼

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

<iframe src="https://willitrunai.com/embed/hf-mradermacher--codestral-21b-pruned-i1-gguf-on-m4-pro-48gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview: