Can I Run Cogito v1 on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM)?

Written by Jakub Rusinowski · Last updated March 20, 2026

Yes, but it is tight

Yes, but it is tight — Cogito v1 32B at Q6_K needs about 29.2 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving only ~2.8 GB before the runtime starts swapping. Expect ~44.3 tok/s (estimated), with room for about 16,384 tokens of context.

Confidence: medium · Recommended quantization: Q6_K · Estimated speed: ~44.3 tok/s

RTX 5090 Desktop (32 GB VRAM, 64 GB RAM) — what it gives a model

Usable memory for models	32 GB
Memory bandwidth	1792 GB/s
Form factor	Desktop
Operating system	Windows or Linux
Memory upgradeable	Yes

Cogito v1 on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM): memory by quantization

Quant	Memory needed	Fits 32 GB?	Max context	Est. speed	Download
F16	66.9 GB	✗ No	—	—	64 GB
Q8_0	36.9 GB	✗ No	—	—	34 GB
Q6_K	29.2 GB	✓ Yes	16K	~44.3 tok/s	26.2 GB
Q5_K_M	25.6 GB	✓ Yes	16K	~50.2 tok/s	22.7 GB
Q4_K_M	22.3 GB	✓ Yes	32K	~57.4 tok/s	19.3 GB
Q3_K_M	16.6 GB	✓ Yes	64K	~75.8 tok/s	13.6 GB
Q2_K	13.5 GB	✓ Yes	64K	~92 tok/s	10.5 GB

Which Cogito v1 sizes fit

Variant	Needs at Q4_K_M	Fits?	Est. speed
Cogito v1 70B	45.7 GB	✗ Too large	—
Cogito v1 32B	22.3 GB	✓ Fits	~57.4 tok/s
Cogito v1 14B	10.9 GB	✓ Fits	~109.5 tok/s
Cogito v1 8B	6.7 GB	✓ Fits	~160.3 tok/s
Cogito v1 3B	3.6 GB	✓ Fits	~253.6 tok/s

What to watch out for

1 larger variant of Cogito v1 does not fit and would need CPU offload or different hardware.

RTX 5090 desktop limitations

575 W board power — budget for a 1000 W+ PSU and the heat it puts into the room.
Models larger than 32 GB must offload to system RAM, which costs roughly an order of magnitude in speed.

Recommended setup

Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving

How these numbers are calculated

Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
32 GB of VRAM on the NVIDIA GeForce RTX 5090 at 1792 GB/s.
64 GB of system RAM available for CPU offload when a model exceeds VRAM.
Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
KV cache is computed from this model's published attention configuration.

FAQ