Models › Qwen3.6-35B-A3B

Qwen3.6-35B-A3B: VRAM needs and speed by GPU

At UD-Q4_K_M, Qwen3.6-35B-A3B needs about 23.8 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.

34.66B parameters (mixture-of-experts, ~3.0B active) · license: apache-2.0 · 262,144-token context · GGUF published Apr 16, 2026 · GGUF files on Hugging Face

Memory needed per quantization

QuantFile sizeNeeds (estimate)Smallest card it fits
UD-IQ1_M10.0 GB11.7 GB16 GB cards
UD-IQ2_XXS10.8 GB12.4 GB16 GB cards
UD-IQ2_M11.5 GB13.2 GB16 GB cards
UD-Q2_K_XL12.3 GB14.0 GB16 GB cards
UD-IQ3_XXS13.2 GB14.9 GB16 GB cards
UD-IQ3_S13.7 GB15.3 GB16 GB cards
UD-Q3_K_S15.4 GB17.0 GB24 GB cards
UD-Q3_K_M16.6 GB18.3 GB24 GB cards
UD-Q3_K_XL16.8 GB18.5 GB24 GB cards
UD-IQ4_XS17.7 GB19.4 GB24 GB cards
UD-Q4_K_S20.9 GB22.6 GB24 GB cards
MXFP4_MOE21.7 GB23.4 GB24 GB cards
UD-Q4_K_M22.1 GB23.8 GB32 GB cards
UD-Q4_K_XL22.4 GB24.0 GB32 GB cards
UD-Q5_K_S24.9 GB26.6 GB32 GB cards
UD-Q5_K_M26.5 GB28.1 GB32 GB cards
UD-Q5_K_XL26.6 GB28.3 GB32 GB cards
UD-Q6_K29.3 GB31.0 GB32 GB cards
UD-Q6_K_XL31.8 GB33.5 GB48 GB cards
Q8_036.9 GB38.6 GB48 GB cards
UD-IQ4_NL37.5 GB39.2 GB48 GB cards
UD-Q8_K_XL38.5 GB40.1 GB48 GB cards
BF1669.4 GB71.0 GB96 GB cards

Expected speed on popular hardware

HardwareBest quant that fitsExpected tok/s
NVIDIA RTX 3060Does not fit
NVIDIA RTX 4060 TiDoes not fit
NVIDIA RTX 3090UD-Q4_K_S183 estimate
NVIDIA RTX 4090UD-Q4_K_S203 estimate
NVIDIA RTX 5090UD-Q6_K229 estimate
AMD Radeon RX 7900 XTXUD-Q4_K_S204 estimate
Apple M4 Pro (20-core GPU) (64 GB)Q8_036.6 estimate
Apple M3 Max (40-core GPU) (64 GB)Q8_053.5 estimate
AMD Ryzen AI Max+ 395 (Radeon 8060S)Q8_032.9 estimate

Check your own card

Too big for your card?

Run Qwen3.6-35B-A3B on a rented 24-80 GB GPU by the hour.

Rent a bigger GPU on RunPod

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