24 GB VRAM (+4)936 GB/s (+576)
C
Raises estimated decode speed by about 133%.51.1 tok/s decode
Raises estimated decode speed by about 133%.
~$1,499 MSRP
Can Codestral 21B Pruned i1 run on RTX 4000 Ada 20GB?
YES — Tight Fit
C49Usable
○Estimated from fit model
Codestral 21B Pruned i1 needs ~18.5 GB VRAM. RTX 4000 Ada 20GB has 20.0 GB. With Q4_K_M quantization, expect ~22 tok/s.
Runtime: OllamaCapacity: TightBandwidth: LowStack: BasicBottleneck: Balanced
Share:
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.
Select quantization to explore
Q4_K_M (Medium quality) — 18.5 GB, 21.9 tok/s, Tight fit
18.5 GB required20.0 GB available
93% VRAM used
Fit status
Tight fit
Decode
21.9 tok/s
TTFT
8832 ms
Safe context
26K
Memory
18.5 GB / 20.0 GB
Memory breakdown
Weights12.8 GB
KV Cache2.5 GB
Runtime1.2 GB
Headroom2.0 GB
See how fast it feels
See how fast it feelsCodestral 21B Pruned i1 on RTX 4000 Ada 20GB
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: 21.9 tok/s decode · 8.8s TTFT (warm) · 55 tok/s prefill
What limits this setup
This setup is broadly balanced for this model.
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
Buy headroom, not only minimum fit
A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.
Performance by workload
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | C | Tight fit | 21.9 tok/s | 4817 ms | 26K |
| Coding | C | Tight fit | 21.9 tok/s | 8832 ms | 26K |
| Agentic Coding | C | Runs with offload (needs ~0.6 GB host RAM) | 14.9 tok/s | 18852 ms | 26K |
| Reasoning | C | Tight fit | 21.9 tok/s | 10438 ms | 26K |
| RAG | C | Runs with offload (needs ~0.6 GB host RAM) | 14.9 tok/s | 23565 ms | 26K |
Quantization options
How Codestral 21B Pruned i1 (21B params) fits at each quantization level on RTX 4000 Ada 20GB (20.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 8.2 GB | Low | C49 |
Q3_K_S | 3 | 10.3 GB | Low | C50 |
NVFP4 | 4 | 11.8 GB | Medium | C50 |
Q4_K_M | 4 | 12.8 GB | Medium | C50 |
Q5_K_MBest for your GPU | 5 | 15.1 GB | High | C49 |
Q6_K | 6 | 17.2 GB | High | F0 |
Q8_0 | 8 | 22.5 GB | Very High | F0 |
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 startUpgrade options