Skip to content
RigCheck

Models / Qwen

Can an RTX 4090 run Qwen3.6 35B-A3B?

MoE with 35B stored and 3B active. Faster than the parameter total suggests; memory still holds every expert.

Stored parameters
35B
Active parameters
3B
Q4 total (estimate)
22.1 GB
License
See model card

Model card. Qwen's card: 35B total parameters, 3B activated, 262,144-token native context.

Compatibility

HardwareQ4Q8FP16
M5 Ultra 256GB
Apple · 256 GB unified
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
Ryzen AI Max+ 395 128GB
AMD · 128 GB unified
70.9 GB
Fits
estimate · KV at 8k
No measured run
M5 Max 128GB
Apple · 128 GB unified
22.1 GB
Fits
estimate · KV at 8k
Measured · 107 tok/s median
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
4× RTX 3090
NVIDIA · 96 GB VRAM
22.1 GB
Fits
estimate · KV at 8k
Measured · 149 tok/s median
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
Ryzen AI Max+ 395 64GB
AMD · 64 GB unified
22.1 GB
Fits
estimate · KV at 8k
Measured · 52.3 tok/s median
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M5 Max 36GB
Apple · 36 GB unified
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
Intel Arc Pro B70
Intel · 32 GB VRAM
22.1 GB
Fits
estimate · KV at 8k
Measured · 120 tok/s median
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M6 32GB
Apple · 32 GB unified
22.1 GB
Fits
estimate · KV at 8k
Measured · 63.2 tok/s median
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 3090
NVIDIA · 24 GB VRAM
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 4090
NVIDIA · 24 GB VRAM
22.1 GB
Tight
estimate · KV at 8k
Measured · 161 tok/s median
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
TITAN RTX
NVIDIA · 24 GB VRAM
22.1 GB
Tight
estimate · KV at 8k
Measured · 113 tok/s median
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
Radeon RX 7900 XTX
AMD · 24 GB VRAM
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M3 Ultra 512GB
Apple · 512 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
2× NVIDIA GB10
NVIDIA · 256 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
M3 Ultra 256GB
Apple · 256 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
M2 Ultra 192GB
Apple · 192 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
NVIDIA GB10
NVIDIA · 128 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
M1 Ultra 128GB
Apple · 128 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
RTX Pro 6000 Blackwell
NVIDIA · 96 GB VRAM
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Fits
estimate · KV at 8k
No measured run
M5 Ultra 96GB
Apple · 96 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Tight
estimate · KV at 8k
No measured run
2× RTX 3090
NVIDIA · 48 GB VRAM
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Fits
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M4 Pro 48GB
Apple · 48 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Tight
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M4 Max 36GB
Apple · 36 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 5090
NVIDIA · 32 GB VRAM
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
Radeon AI Pro R9700
AMD · 32 GB VRAM
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M4 32GB
Apple · 32 GB unified
22.1 GB
Fits
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M4 Pro 24GB
Apple · 24 GB unified
22.1 GB
Tight
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M3 Pro 18GB
Apple · 18 GB unified
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 4060 Ti 16GB
NVIDIA · 16 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 5060 Ti 16GB
NVIDIA · 16 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 5070 Ti
NVIDIA · 16 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
M4 16GB
Apple · 16 GB unified
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 3060 12GB
NVIDIA · 12 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 3080 Ti
NVIDIA · 12 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 4070 Ti
NVIDIA · 12 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run
RTX 2060 Super
NVIDIA · 8 GB VRAM
22.1 GB
Over
estimate · KV at 8k
No measured run
38.1 GB
Over
estimate · KV at 8k
No measured run
70.9 GB
Over
estimate · KV at 8k
No measured run

Measured runs

HardwareQuanttok/sRuntimeSource
M5 Max 128GB4-bit MLX (mlx-serve build)271 · speculativemlx-serve 26.10.2 + MTPSource · checked
RTX Pro 6000 BlackwellNVFP4261vLLM, b12x FP4 MoE kernelSource · checked · prefill 26600
RTX 5090llama-bench tg128 (quant per result)258llama.cppSource · checked
M5 Ultra 96GBOllama qwen3.6:35b-a3b default tag (23 GB)246Ollama 0.40.13 (llamacpp runner)Source · checked
M5 Max 128GB4-bit MLX240 · speculativeRapid-MLX 0.16.1 + MTPSource · checked
4× RTX 3090AWQ-INT4 + MTP-3 (TP=2)225–231 · speculativevLLMSource · checked
TITAN RTXUnsloth UD-Q4_K_M with embedded MTP head (21.1 GiB), q8_0 KV, 65k context176.4 · speculativellama.cpp b11222 (CUDA), llama-server, MTP draft 2Source · checked
M5 Max 36GB4-bit MLX173.4 · speculativeRapid-MLX 0.16.1 (MTP)Source · checked
RTX 3090Q4_K_M172.4llama.cpp CUDA b11804Source · checked · prefill 2763
RTX 3090Q4_K_M162.24llama.cpp v0.7.0 llama-bench (CUDA), fully on GPUSource · checked · prefill 4362
RTX 4090Q4_K_M160.6llama.cpp b8931Source · checked · prefill 4589
M5 Ultra 256GBMLX affine 4-bit157MLXSource · checked · prefill 8700
4× RTX 3090AWQ-INT4 (TP=2)149vLLMSource · checked
RTX 3090llama-bench tg128 (quant per result)146–153llama.cppSource · checked
M5 Max 36GB4-bit MLX145.8oMLX 0.7.0Source · checked
M5 Ultra 256GBMLX NVFP4142MLXSource · checked
RTX 3090Q4 GGUF, all layers on GPU, ~90K ctx139.6llama.cpp CUDASource · checked · prefill 3360
Radeon RX 7900 XTXUD-Q4_K_XL + MTP (draft-n-max 4), ~16K context135.2 · speculativellama.cpp ROCm (llama-server)Source · checked · prefill 1817
M5 Ultra 256GBGGUF Unsloth UD-Q4_K_XL134llama.cppSource · checked
M5 Max 36GB4-bit MLX124.5LM Studio (MLX)Source · checked
Intel Arc Pro B70GPTQ Int4120vLLM XPU b70 image, no MTPSource · checked · prefill 9300
M5 Ultra 256GBMLX MXFP8118MLXSource · checked
4× RTX 3090IQ4_XS (1 GPU)115–133llama.cppSource · checked
TITAN RTXUnsloth UD-Q4_K_M with embedded MTP head (21.1 GiB), q8_0 KV, 65k context113.4llama.cpp b11222 (CUDA)Source · checked · prefill 2180
NVIDIA GB10NVFP4113 · speculativevLLM 0.30 + MTPSource · checked · prefill 5860
Radeon RX 7900 XTXUD-Q4_K_XL + MTP, ~64K context108.2 · speculativellama.cpp ROCm (llama-server)Source · checked · prefill 1289
M5 Max 128GB4-bit MLX107mlx_lm.server 0.32.2Source · checked
NVIDIA GB10NVFP4 + MTP (k=2)102.3 · speculativevLLM 0.31.0, 128K ctx, 1 userSource · checked
NVIDIA GB10NVFP4 + MTP95.1 · speculativevLLM, 1 streamSource · checked
Ryzen AI Max+ 395 128GBQ4_K_M (Unsloth)93 · speculativellama.cpp Vulkan + MTPSource · checked
Radeon RX 7900 XTXUD-Q4_K_XL, llama-bench 2K, f16 KV87.2llama.cpp ROCmSource · checked · prefill 2724
NVIDIA GB10NVFP486 · speculativevLLM 0.30 + MTPSource · checked · prefill 5860
Ryzen AI Max+ 395 128GBUnsloth UD-Q4_K_M with embedded MTP head (21.1 GiB), q8_0 KV, 65k context81.9 · speculativellama.cpp (Vulkan), llama-server, MTP draft 3Source · checked
Radeon RX 7900 XTXUD-Q4_K_XL, ~16K real-text context, q4_0 KV79.6llama.cpp ROCm (llama-server)Source · checked · prefill 1918
NVIDIA GB10NVFP4, no speculative decoding78.1vLLM 0.31.0, 128K ctx, 1 userSource · checked
RTX 3090Q4_K_M, 20 layers of experts in system RAM (-ncmoe 20)73.59llama.cpp v0.7.0 llama-server + --moe-cache-mib 1866Source · checked · prefill 301
M6 32GBMLX affine 4-bit (group 64) + MTP head and DFlash2 drafter69.3 · speculativeKapitsa 0.2.4 (KFlash tree speculative decoding)Source · checked
Ryzen AI Max+ 395 64GBUD-Q4_K_M (22.7 GB) + MTP draft N=268.03 · speculativellama.cpp b11065 (ROCm)Source · checked · prefill 1240
Radeon RX 7900 XTXUD-Q4_K_XL, ~64K context66.0llama.cpp ROCm (llama-server)Source · checked · prefill 1368
M6 32GBMLX 4-bit63.2LM Studio (MLX)Source · checked
Ryzen AI Max+ 395 128GBQ4_K_M (Unsloth)62llama.cpp VulkanSource · checked · prefill 1380
Ryzen AI Max+ 395 128GBUnsloth UD-Q4_K_XL (22.4 GB)61.8llama.cpp b11401 (Windows x64 Vulkan, unmodified)Source · checked
Ryzen AI Max+ 395 128GBQ8_0 + MTP (draft-n-max 3), coding workload60.92 · speculativellama.cpp ROCmSource · checked
Ryzen AI Max+ 395 128GBQ4_K_M59.4llama.cpp b11213 VulkanSource · checked
Ryzen AI Max+ 395 128GBUnsloth UD-Q4_K_M with embedded MTP head (21.1 GiB), q8_0 KV, 65k context57.4llama.cpp (Vulkan), llama-serverSource · checked
Ryzen AI Max+ 395 128GB4-bit54.4Ollama 0.34.0Source · checked
Ryzen AI Max+ 395 128GBQ8_0, llama-bench tg12854.22llama.cpp VulkanSource · checked · prefill 911
Ryzen AI Max+ 395 128GBQ8_054llama.cpp VulkanSource · checked · prefill 1370
Ryzen AI Max+ 395 64GBUD-Q4_K_M (22.1 GB)52.3llama.cpp b11669 (Vulkan)Source · checked · prefill 1006
M6 32GBSplash package (incoai/Qwen3.6-35B-A3B-Splash)51.5 · speculativeLM Studio Splash backendSource · checked
Ryzen AI Max+ 395 128GBQ4_K_M (Unsloth)51llama.cpp ROCmSource · checked · prefill 1310
Ryzen AI Max+ 395 128GBQ8_0, llama-bench tg12847.47llama.cpp ROCmSource · checked · prefill 925
Radeon RX 7900 XTXUD-Q4_K_XL, ~245K context, q4_0 KV38.3llama.cpp ROCm (llama-server)Source · checked · prefill 641
Ryzen AI Max+ 395 64GBUD-Q4_K_M (22.7 GB) + MTP draft N=221.66 · speculativellama.cpp b11065 (ROCm)Source · checked