Can the RTX 5090 D run Qwen3-VL-8B-Instruct?

Yes. Qwen3-VL-8B-Instruct fits the RTX 5090 D's 32 GB at Q4_K_M (about 6.8 GB needed), running at an estimated 205 tok/s. There's room for up to FP16 if you want higher quality.

VRAM
32 GB
Needs at Q4
6.8 GB
Headroom
25.2 GB
Speed (Q4_K_M)
~205 tok/s

Every quantization

Qwen3-VL-8B-Instruct (8.8B) on 32 GB

QuantWeightsNeedsFitEst. speed
Q4_K_M5 GB6.8 GBFits easily~205 tok/s
Q5_K_M6 GB7.8 GBFits easily~171 tok/s
Q8_09 GB10.8 GBFits easily~114 tok/s
FP1618 GB19.8 GBFits~57 tok/s

Estimates. Needs = weights + 1.8 GB for KV cache and runtime at modest context. Speed is single-stream decode scaled from the RTX 5090 D's 1790 GB/s bandwidth (Estimated from memory bandwidth (llama.cpp decode)). Long contexts need more memory and run slower.

Run it

Copy, paste, go

llama-server -hf Qwen/Qwen3-VL-8B-Instruct-GGUF -ngl 99

Ollama pulls a Q4 build by default. Pick a larger quant only if the table above says it fits. Full Qwen3-VL-8B-Instruct details →

Compare

Same model, other hardware - and other models on this card

Qwen3-VL-8B-Instruct on other hardware

Also runs on the RTX 5090 D

Frequently asked

Quick answers

Can the GeForce RTX 5090 D run Qwen3-VL-8B-Instruct?

Yes. Qwen3-VL-8B-Instruct fits the RTX 5090 D's 32 GB at Q4_K_M (about 6.8 GB needed), running at an estimated 205 tok/s. There's room for up to FP16 if you want higher quality.

How much VRAM does Qwen3-VL-8B-Instruct need?

About 5 GB of weights at Q4_K_M, 9 GB at Q8_0 and 18 GB at FP16. Add roughly 1.8 GB for the KV cache and runtime at modest context lengths.

How fast is Qwen3-VL-8B-Instruct on the RTX 5090 D?

Roughly 205 tokens per second at Q4_K_M (single user, llama.cpp-style decode). This is an estimate from the RTX 5090 D's 1790 GB/s memory bandwidth.