How fast does Qwen3.6 35B-A3B run on the NVIDIA RTX 3090?

Qwen3.6 35B-A3B runs at about 158 tokens/s on the NVIDIA RTX 3090 (measured, UD-Q4_K_M, CUDA, llama-bench). Based on 1 measured run(s), each linked to its public source.

Does Qwen3.6 35B-A3B UD-Q4_K_M fit in the NVIDIA RTX 3090's memory?

Estimate: the UD-Q4_K_M file is 22.1 GB and needs about 23.8 GB with an ~8K context; the NVIDIA RTX 3090 has about 23.5 GB usable, so it does not fully fit (part of it runs on the CPU).

Qwen3.6 35B-A3B › NVIDIA RTX 3090

Qwen3.6 35B-A3B on the NVIDIA RTX 3090

Qwen3.6 35B-A3B runs at about 158 tokens/s on the NVIDIA RTX 3090 (measured, UD-Q4_K_M, CUDA, llama-bench).

Best measured: 158 tokens/s generation (UD-Q4_K_M, CUDA). The source gives no prompt-processing figure.

For comparison, our size-scaling estimate for this combination is 173 tok/s.

Every run

Quant	Backend	Generate tok/s	Prompt tok/s	Test	Build / by / notes	Source
UD-Q4_K_M	CUDA	158	–	tg128	Headless, whole model on the card (no expert offload), default KV cache.	InsiderLLM Qwen 3.6 35B guide

tg128 = generating 128 tokens; pp512 = reading a 512-token prompt (llama-bench). LocalScore averages several prompt and output lengths, so its numbers run lower than tg128. "decode, llama-server" = single-stream generation speed measured against a running llama-server (no speculative decoding).

Check prices

Amazon Newegg

As an Amazon Associate I earn from qualifying purchases.

No card yet?

Try it on a rented GPU (RunPod)

Affiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.

All results for the NVIDIA RTX 3090 · Qwen3.6 35B-A3B on other hardware

Embed this result

HTML

<a href="https://rigbench.pages.dev/model/qwen3-6-35b-a3b/on/nvidia-rtx-3090/"><img src="https://rigbench.pages.dev/badge/qwen3-6-35b-a3b--nvidia-rtx-3090.svg" alt="Qwen3.6 35B-A3B · NVIDIA RTX 3090: 158 tok/s (UD-Q4_K_M, measured)"></a>

Markdown (GitHub README, forums)

[![Qwen3.6 35B-A3B · NVIDIA RTX 3090: 158 tok/s (UD-Q4_K_M, measured)](https://rigbench.pages.dev/badge/qwen3-6-35b-a3b--nvidia-rtx-3090.svg)](https://rigbench.pages.dev/model/qwen3-6-35b-a3b/on/nvidia-rtx-3090/)