Gemma Models — Hardware Requirements
17 Gemma models from Google and the community, from the smallest that runs in 1.2 GB of VRAM at Q4 up to 27.4B parameters. Every row links to full quantization tables and GPU compatibility.
All Gemma Models by Size
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Functiongemma 270M IT | 268M | 1.2 GB | — | ||
| T5gemma S S Prefixlm | 313M | 1.3 GB | — | ||
| T5gemma B B Ul2 IT | 591M | 1.6 GB | — | ||
| Vaultgemma 1B | 1.0B | 2.1 GB | — | ||
| T5gemma L L Ul2 IT | 1.2B | 2.3 GB | — | ||
| Gemma 1.1 2B IT | 2.5B | 3.5 GB | — | ||
| Medgemma 1.5 4B IT | 4.3B | 5 GB | — | ||
| Gemma 7B | 8.5B | 8.2 GB | — | ||
| Codegemma 7B IT | 8.5B | 8.2 GB | — | ||
| Gemma 7B IT | 8.5B | 8.2 GB | — | ||
| Shieldgemma 9B | 9.2B | 8.8 GB | — | ||
| Gemma SEA LION v3 9B IT | 9.2B | 7.7 GB | 8K | ||
| Vllm Translategemma 12B IT | 13.2B | 9.8 GB | 131K | ||
| Medgemma 27B Text IT | 27.0B | 21.4 GB | — | ||
| Medgemma 27B IT | 27.4B | 21.7 GB | — |
Community variants and fine-tunes(3)
| Model | Params | Runs from (Q4) | Context | Publisher | Quant downloads |
|---|---|---|---|---|---|
| Functiongemma 270M Ft Mobile Actions | 270M | 1.2 GB | — | ||
| Turkish Gemma 9B T1 | 9.2B | 7.7 GB | 8K | ||
| Turkish Gemma 9B v0.1 | 9.2B | 7.7 GB | 8K |
Frequently Asked Questions
- How much VRAM do I need to run a Gemma model?
- The smallest Gemma model, Functiongemma 270M IT, runs from 1.2 GB of VRAM at Q4_K_M (smaller low-bit quantizations start at 0.6 GB with a noticeable quality loss). Larger family members need proportionally more — see the table above for every model.
- Which Gemma models can I run on a 16 GB GPU?
- 16 of 18 Gemma models fit in 16 GB of VRAM at Q4_K_M, including Medgemma 1.5 4B IT, Gemma 7B, Codegemma 7B IT.
- What is the most popular Gemma model to run locally?
- Medgemma 27B Text IT is the most downloaded Gemma model in local-friendly quantized formats. It runs from 21.4 GB of VRAM at Q4_K_M.