Can I run it on my PC?
Answers for the open models people ask about most, on every consumer GPU and Mac we track. For any other model, use the VRAM calculator or the model list.
Can I run gpt-oss-120b?
gpt-oss 120B: 59.5 GB at 4-bit
No single consumer GPU in VRAM · Mac: Mac M2 Max 96GB
Can I run gpt-oss-20b?
gpt-oss 20B: 11.8 GB at 4-bit
From a Arc A770 16GB ($349) · Mac: Mac M4 Pro 24GB
Can I run Llama 70B?
5 models · Llama 3.3 70B Instruct: 43.3 GB at 4-bit
No single consumer GPU in VRAM · Mac: Mac M2 Max 96GB
Can I run Qwen3?
21 models · Qwen3 32B: 21.1 GB at 4-bit
From a RTX 5090 ($1999) · Mac: Mac M3 Pro 36GB
Can I run DeepSeek R1?
7 models · DeepSeek R1: 378 GB at 4-bit
No single consumer GPU in VRAM
New releases
Added automatically when a new model picks up traction on Hugging Face. See new this week.
Can I run SenseNova Loop 9B?
6.4 GB at 4-bit
From a Arc B580 ($249) · Mac: Mac M4 16GB
Can I run KORMo Web 11B?
8.3 GB at 4-bit
From a Arc B580 ($249) · Mac: Mac M4 16GB
Can I run SingVL S1 4B?
3.6 GB at 4-bit
From a Arc B580 ($249) · Mac: Mac M4 16GB
Can I run Nex N2.6?
188 GB at 4-bit
No single consumer GPU in VRAM · Mac: Mac M3 Ultra 512GB
Can I run MiniMax H3 Image PE?
16.8 GB at 4-bit
From a RX 7900 XT ($899) · Mac: Mac M1 Pro 32GB
Can I run Arete 27B?
17.4 GB at 4-bit
From a RX 7900 XTX ($999) · Mac: Mac M1 Pro 32GB
Can I run Inkling 2 Small?
346 GB at 4-bit
No single consumer GPU in VRAM
Can I run LongCat 2.5?
1014 GB at 4-bit
No single consumer GPU in VRAM