Models › Qwen3.8-27B

Qwen3.8-27B: VRAM needs and speed by GPU

At UD-Q4_K_M, Qwen3.8-27B needs about 17.9 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.

27.32B parameters · license: apache-2.0 · 262,144-token context · GGUF published Aug 13, 2026 · GGUF files on Hugging Face

Memory needed per quantization

QuantFile sizeNeeds (estimate)Smallest card it fits
UD-IQ1_S6.2 GB7.7 GB12 GB cards
UD-IQ1_M6.7 GB8.2 GB12 GB cards
UD-IQ2_XXS7.3 GB8.7 GB12 GB cards
UD-IQ2_S8.4 GB9.9 GB12 GB cards
UD-Q2_K_XL9.8 GB11.3 GB12 GB cards
UD-IQ3_XXS10.9 GB12.4 GB16 GB cards
UD-IQ3_S12.0 GB13.5 GB16 GB cards
UD-Q3_K_XL13.1 GB14.6 GB16 GB cards
UD-IQ4_XS14.3 GB15.7 GB24 GB cards
UD-Q4_K_S15.4 GB16.8 GB24 GB cards
Q4_016.1 GB17.5 GB24 GB cards
UD-Q4_K_M16.5 GB17.9 GB24 GB cards
Q4_117.5 GB19.0 GB24 GB cards
UD-Q4_K_XL17.6 GB19.0 GB24 GB cards
UD-Q5_K_S18.7 GB20.1 GB24 GB cards
UD-Q5_K_M19.8 GB21.3 GB24 GB cards
UD-Q5_K_XL20.9 GB22.4 GB24 GB cards
UD-Q6_K22.0 GB23.5 GB24 GB cards
UD-Q6_K_M23.1 GB24.6 GB32 GB cards
UD-Q6_K_L24.2 GB25.7 GB32 GB cards
UD-Q6_K_XL25.3 GB26.8 GB32 GB cards
UD-Q8_K_L28.0 GB29.5 GB32 GB cards
Q8_029.0 GB30.5 GB32 GB cards
UD-Q8_K_XL31.5 GB32.9 GB48 GB cards
BF1654.7 GB56.1 GB96 GB cards

Expected speed on popular hardware

HardwareBest quant that fitsExpected tok/s
NVIDIA RTX 3060UD-IQ2_XXS42.4 estimate
NVIDIA RTX 4060 TiDoes not fit
NVIDIA RTX 3090UD-Q6_K29.8 estimate
NVIDIA RTX 4090UD-Q6_K33.1 estimate
NVIDIA RTX 5090Q8_039.5 estimate
AMD Radeon RX 7900 XTXUD-Q6_K33.2 estimate
Apple M4 Pro (20-core GPU) (64 GB)Q8_08.0 estimate
Apple M3 Max (40-core GPU) (64 GB)Q8_011.6 estimate
AMD Ryzen AI Max+ 395 (Radeon 8060S)Q8_07.2 estimate

Check your own card

Too big for your card?

Run Qwen3.8-27B 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.