Open-weight LLMs you can run yourself › Gemma 4 26B A4B
Updated · public data
Google · Open weights
Gemma 4 26B A4B is an open-weight LLM from Google with 27B parameters. It needs about 16 GB of memory at 4-bit — smallest setups that run it: PC with a GPU 24 GB, PC or laptop, no GPU 32 GB, Mac 32 GB.
| Parameters | 27B |
| Memory to run | ~16 GB (at 4-bit, rough estimate) |
| Downloads | 11.2M (last 30 days on Hugging Face) |
| Likes | 1.7K |
| License | Apache 2.0 |
| First uploaded | Mar 11, 2026 |
| Hosted from | $0.09 (per 1M input tokens · cheapest on OpenRouter) |
| Phone | iPhone · Android: too big |
| PC with a GPU | 8 GB: too big · 12 GB: too big · 16 GB: too big · 24 GB: fits · 32 GB: fits |
| PC or laptop, no GPU | 16 GB: too big · 32 GB: fits · 64 GB: fits |
| Mac | 16 GB: too big · 32 GB: fits · 64 GB: fits · 128 GB: fits |
huggingface.co/google/gemma-4-26B-A4B-it
Data: Hugging Face (official accounts), Oct 16, 2026 · memory is a rough estimate (4-bit for LLMs, 8-bit for image and video, plus 10%) · hosted price: OpenRouter