Qwen3.6-35B-A3B

Qwen family · 36.0B params

17recorded runs

969.4best tok/s

9verified

Evidence: 9 sourced runs · 8 awaiting review · 2 independent sources · first recorded Aug 5, 2026 · last recorded Oct 14, 2026. What these labels mean.

What the Qwen3.6-35B-A3B numbers mean

The fastest verified result on our board is 969.4 output tok/s on an Radeon AI Pro R9700 at MQ4R with a 4096-token context. That is about 1.0 ms per generated token — 500 tokens takes roughly 0.5 seconds. Across the 17 records we hold for this model the range is 4.3–969.4 tok/s on 12 GPUs, because quantisation, context length and the engine build all move the number.

How much VRAM does Qwen3.6-35B-A3B need?

Qwen3.6-35B-A3B has 36.0B 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)	24.1 GB	32 GB
Q8_0 (8-bit)	44.8 GB	64 GB
FP16 (16-bit)	83.6 GB	141 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
Ornith-1.5-35B-A3B	36B	20.2 GB
Ornith-1.5-35B-A3B-GPTQ-Int4-sym-G128-MTP-BF16-MixedCal-v2	36B	20.2 GB
Ornith-1.5-35B-A3B-MTP	36B	20.2 GB
Qwen3.6-35B-A3B-abliterated-v4	36B	20.2 GB
Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved	36B	20.2 GB
SIQ-1-35B-B70-Turbo	36B	20.2 GB

Ranked by distance from Qwen3.6-35B-A3B's parameter count. Full benchmark board.

All benchmarks Methodology

GPU	Model	Quant	Backend	Ctx	tok/s	prompt tok/s	VRAM GB	Source
Radeon AI Pro R9700 · 32 GB	Qwen3.6-35B-A3B	MQ4R	hipfire 0.3.0+b0bcc3f91506	4096	969.4 ✓	—	—	src
RTX PRO 6000 Blackwell · 96 GB	Qwen3.6-35B-A3B	NVFP4	vllm 0.26.0	262144	330.2 ✓	11518.9	—	src
Intel Arc Pro B70 · 32 GB	Qwen3.6-35B-A3B	GPTQ-4bit	vllm vLLM 0.26.1.dev0 XPU, built from v0.…	32768	248.3 ✓	12400.0	—	src
Mac Studio (M4 Max) · 64 GB	Qwen3.6-35B-A3B	Q4_K_XL	llama.cpp 781dbc5	1000	226.9 ✓	—	—	src
Mac Studio (M4 Max) · 64 GB	Qwen3.6-35B-A3B	Q4_K_XL	llama.cpp 781dbc5	4000	220.2 ✓	—	—	src
Mac Studio (M4 Max) · 64 GB	Qwen3.6-35B-A3B	Q4_K_XL	llama.cpp 781dbc5	400	207.6 ✓	—	—	src
RX 7900 XTX · 24 GB	Qwen3.6-35B-A3B	MQ4R	hipfire 0.3.0+32d27a87d7f0	4096	150.0 ✓	—	—	src
CMP 170HX · 64 GB	Qwen3.6-35B-A3B	UD-Q8_K_XL	llama.cpp 0.3.0-dev+b10705	131338	87.2 ✓	1322.2	—	src
Tesla V100 · 32 GB	Qwen3.6-35B-A3B	Q4_K_M	llama.cpp e031d95	262144	4.3 ✓	322.5	28.9	src
RTX 5090 · 32 GB	Qwen3.6-35B-A3B	Q4_K_M	lmstudio —	2048	243.1 …	625.2	—	src
RTX 3060 x2 · 12 GB	Qwen3.6-35B-A3B	UD-Q4_K_XL	llama.cpp —	90112	129.4 …	—	—	src
Radeon AI Pro R9700 · 32 GB	Qwen3.6-35B-A3B	Q4_K_XL	llama.cpp —	2048	78.6 …	2883.3	—	src
Multi-GPU x3 · 72 GB	Qwen3.6-35B-A3B	APEX-I-Balanced	llama.cpp 81 (15586e2)	2097152	63.3 …	—	65.7	src
CMP 170HX · 64 GB	Qwen3.6-35B-A3B	FP8	vllm 0.28.0	131335	62.0 …	4982.8	—	src
T4 x3 · 48 GB	Qwen3.6-35B-A3B	INT4_W4A16	vllm 0.28.0	200000	46.6 …	353.7	38.5	src
Intel Arc Pro B70 · 32 GB	Qwen3.6-35B-A3B	Q4_0	llama.cpp —	2048	45.6 …	1376.4	—	src
Intel Arc Pro B60 · 24 GB	Qwen3.6-35B-A3B	Q4_0	llama.cpp —	2048	37.4 …	908.0	—	src