Model Anatomy See what's inside open-weight AI models

Can I run Huihui-Qwen3.8-27B-abliterated-GGUF on an NVIDIA B300?

Yes

Its BF16 file (54.7 GB) fits in the 259 GB an NVIDIA B300 can use, with room for ≈ 256K tokens of context.

Full analysis of Huihui-Qwen3.8-27B-abliterated-GGUF Everything that runs on NVIDIA B300 Get the BF16 file on Hugging Face
ollama run hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:BF16

Files and context on NVIDIA B300

Each file of Huihui-Qwen3.8-27B-abliterated-GGUF (and of GGUF versions the site has analyzed), and how much context fits next to it, with the context cache in 16-bit or 8-bit.

FileSizeOn NVIDIA B300 Context (16-bit)Context (8-bit)
BF1654.7 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-bf1654.7 GBFits256K256K
F1653.8 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Q8_0_L38.8 GBFits256K256K
UD-Q8_K_XL31.5 GBFits256K256K
Q8_029.0 GBFits256K256K
Q8_K_L27.3 GBFits256K256K
Q6_K_L24.9 GBFits256K256K
UD-Q6_K_XL24.8 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-UD-DW-Q6_K_L23.9 GBFits256K256K
Q5_K_L22.9 GBFits256K256K
Q6_K22.4 GBFits256K256K
Q4_K_L20.9 GBFits256K256K
UD-Q5_K_XL20.7 GBFits256K256K
Q5_K19.5 GBFits256K256K
Q3_K_L18.7 GBFits256K256K
UD-Q4_K_XL17.4 GBFits256K256K
Q2_K_L17.2 GBFits256K256K
Q4_K16.8 GBFits256K256K
Q4_K_M16.6 GBFits256K256K
Q4_K_S15.6 GBFits256K256K
UD-IQ4_XS14.4 GBFits256K256K
Q3_K13.5 GBFits256K256K
UD-Q3_K_XL13.3 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-UD-DW-IQ3_S12.3 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S-mtp12.1 GBFits256K256K
IQ3_S12.1 GBFits256K256K
UD-IQ3_S12.0 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_S11.8 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-GSQ-RCO-IQ3_S11.8 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-UD-DW-IQ3_XXS11.4 GBFits256K256K
UD-IQ3_XXS11.0 GBFits256K256K
Q2_K10.9 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_XXS-mtp10.4 GBFits256K256K
IQ3_XXS10.4 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ3_XXS10.1 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-GSQ-RCO-IQ3_XXS10.1 GBFits256K256K
UD-Q2_K_XL10.0 GBFits256K256K
IQ2_S9.61 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ2_S9.26 GBFits256K256K
UD-IQ2_S8.78 GBFits256K256K
IQ2_XS8.76 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-UD-IQ2_S8.42 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Swift-1.5-GSQ-RCO-IQ2_XS8.42 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Underdog-Saluki-Swift-1.5-IQ2-mtp8.25 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Underdog-Saluki-Swift-1.5-IQ27.90 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Ternary-Bonsai-PQ2_07.70 GBFits256K256K
Huihui-Qwen3.8-27B-abliterated-Ternary-Bonsai-PTQ1_06.61 GBFits256K256K

Huihui-Qwen3.8-27B-abliterated-GGUF on other hardware

Popular models that run on NVIDIA B300

Assumes 90% of the card's memory is usable and at least 4K tokens of context for one conversation, with the context cache in 16-bit unless noted. Fits leaves at least 20% of that memory for context; Tight leaves less.