Skip to content
Run LLM Locally

Mistral Small 3.2 24B Instruct on the RTX 4090

With 24 GB of VRAM, the RTX 4090 runs Mistral Small 3.2 24B Instruct at Q4_K_M entirely on the GPU with an 8K context (about 16.1 GB needed). At Q8_0 it does not fit in VRAM.

ShareFacebookXWhatsAppLinkedInEmail

Mistral Small 3.2 24B Instruct

Q4_K_M · 8K context

Excellent

The model needs about 16.1 GB and fits in the 23.3 GB available, with 7.1 GB to spare.

Your hardware
RTX 4090 24 GB · 64 GB RAM
Estimated memory
~16.1 GB estimate
VRAM (usable memory)
23.3 GB
GPU only
Yes
CPU/GPU offload
Not required
Generation speed
46–63 tok/s estimate
Largest context in fast memory
51K tokens
Confidence: low

Based on: memory bandwidth (1,008 GB/s) and model size

Memory breakdown

~16.1 GB (15.5 GB–16.8 GB) needed · estimate

Fits entirely in VRAM · 7.1 GB left

  • Model weights13.3 GB
  • KV cache1.3 GB
  • Runtime overhead0.8 GB
  • Safety margin0.8 GB
Assumptions: context 8K tokens · KV cache f16 · llama.cpp (CUDA) · flash attention on · weights = real file size

EstimateData updated 24 days agoSource: Run LLM Locally engine, Hugging Face config.json + model card

What makes this estimate less reliable

  • no benchmark for this hardware and model yet

Other quantizations on the RTX 4090

Mistral Small 3.2 24B Instruct on RTX 4090 + 64 GB RAM · 8K context
Mistral Small 3.2 24B Instruct Q8_08K~26.6 GB · 5–9.5 tok/s estimateCPU/GPU offload
Mistral Small 3.2 24B Instruct Q6_K8K~21 GB · 36–48 tok/s estimateCompatible
Mistral Small 3.2 24B Instruct Q5_K_M8K~18.5 GB · 40–55 tok/s estimateExcellent
Mistral Small 3.2 24B Instruct Q4_K_M8K~16.1 GB · 46–63 tok/s estimateExcellent
Mistral Small 3.2 24B Instruct IQ4_XS8K~14.6 GB · 51–70 tok/s estimateExcellent
Mistral Small 3.2 24B Instruct Q3_K_M8K~13.3 GB · 56–77 tok/s estimateExcellent

Alternatives

GPUs that run Mistral Small 3.2 24B Instruct Q4_K_M entirely in VRAM

Other models that fit the RTX 4090

Where to buy a RTX 4090

Prices and links follow the market you choose.

GeForce RTX 4090

24 GB GDDR6X · 1,008 GB/s · 450 W

ExcellentRuns Mistral Small 3.2 24B Instruct Q4_K_M entirely in memory

Price not available

Availability unknown

View on amazon.com

Affiliate link

No current offer imported for this market: the link opens a marketplace search.

Frequently asked questions

Can the RTX 4090 run Mistral Small 3.2 24B Instruct?
The model needs about 16.1 GB and fits in the 23.3 GB available, with 7.1 GB to spare.
Which quantization should I use on the RTX 4090?
Q6_K is the highest-precision quantization that fits entirely in VRAM at 8K context. ~21 GB / 23.3 GB.
What is the largest context on the RTX 4090?
At Q4_K_M, about 51K tokens fit entirely in VRAM with an f16 KV cache. A q8_0 KV cache roughly doubles that.
How fast is Mistral Small 3.2 24B Instruct (RTX 4090)?
We estimate 46–63 tok/s (estimate, confidence low). Based on: memory bandwidth (1,008 GB/s) and model size.