Qwen3.6-27B

Qwen family · 28.0B params

11recorded runs

240.8best tok/s

6verified

Evidence: 6 sourced runs · 5 awaiting review · 1 independent source · first recorded May 10, 2026 · last recorded Aug 26, 2026. What these labels mean.

What the Qwen3.6-27B numbers mean

The fastest verified result on our board is 240.8 output tok/s on an RTX 3090 x2 at AutoRound INT4 with a 200000-token context. That is about 4.2 ms per generated token — 500 tokens takes roughly 2.1 seconds. Across the 11 records we hold for this model the range is 4.1–240.8 tok/s on 9 GPUs, because quantisation, context length and the engine build all move the number.

How much VRAM does Qwen3.6-27B need?

Qwen3.6-27B has 28.0B parameters. These estimates use the same formula as the VRAM calculator: weights, plus about 15% for a moderate-context KV cache and runtime overhead, plus a small engine floor.

Quantisation	Estimated VRAM	Smallest card that fits
Q4_K_M (4-bit)	18.9 GB	24 GB
Q8_0 (8-bit)	35.0 GB	48 GB
FP16 (16-bit)	65.2 GB	80 GB

Long contexts and large batches raise the real number — read the smallest card as "possible with a reduced context". Measured throughput is on the benchmark board.

Fastest GPUs for this model (best tok/s)

Models with a similar footprint

Model	Params	Q4_K_M weights
Bonsai-27B	28B	15.8 GB
Godoter-27B	28B	15.8 GB
Huihui-Qwen3.8-27B-abliterated	28B	15.8 GB
Qwen3.5-27B	28B	15.8 GB
Qwen3.6-27B-i1	28B	15.8 GB
Qwen3.6-27B-MTP	28B	15.8 GB

Ranked by distance from Qwen3.6-27B's parameter count. Full benchmark board.

All benchmarks Methodology

GPU	Model	Quant	Backend	Ctx	tok/s	prompt tok/s	VRAM GB	Source
RTX 3090 x2 · 24 GB	Qwen3.6-27B	AutoRound INT4	vllm 0.19.1	200000	240.8 ✓	—	—	src
RTX 5090 · 32 GB	Qwen3.6-27B	NVFP4	vllm 0.25.1	131072	120.9 ✓	—	29.9	src
RTX PRO 6000 Blackwell · 96 GB	Qwen3.6-27B	FP8	vllm 0.26.0	131072	101.4 ✓	—	—	src
RTX PRO 6000 Blackwell · 96 GB	Qwen3.6-27B	NVFP4	vllm 0.26.0	131072	73.0 ✓	—	—	src
RTX 3060 x2 · 12 GB	Qwen3.6-27B	IQ4_NL	llama.cpp b10443	32768	18.9 ✓	339.8	—	src
RTX 5090 · 32 GB	Qwen3.6-27B	UD-Q5_K_XL	llama.cpp from environment.txt	131072	4.1 ✓	—	27.5	src
RX 7900 XT · 20 GB	Qwen3.6-27B	IQ4_XS	vllm —	2048	60.6 …	375.0	—	src
Multi-GPU x3 · 72 GB	Qwen3.6-27B	Q6_K_L	llama.cpp 81 (15586e2)	524288	60.3 …	—	51.4	src
Intel Arc Pro B70 · 32 GB	Qwen3.6-27B	sym_int4	vllm —	2048	55.1 …	215.5	—	src
Tesla P100 x2 · 32 GB	Qwen3.6-27B	Q4_H	llama.cpp —	2048	51.3 …	200.0	—	src
RTX 4070 · 12 GB	Qwen3.6-27B	UD-IQ2_M	llama.cpp —	65536	23.5 …	456.8	11.8	src