Can I run gemma-3n-E2B-it?
gemma-3n-E2B-it needs about 4.8 GB of memory at Q4_K_M, and the cheapest GPU we track that holds it is the Intel Arc B570 (10 GB, $219 list price). 82 of the 82 devices we track run it fully at Q4. These are estimates, not measured benchmarks.
- Q4_K_M needs
- 4.8 GB 3.3 GB weights
- Q8_0 needs
- 7.6 GB 5.8 GB weights
- FP16 needs
- 13.5 GB 10.9 GB weights
82 of 82 tracked GPUs and Macs run it fully at Q4.
Which hardware runs it
Bar: Q4 memory needed vs. usable memory. The tick marks 90%, our comfort line.FitsTightWon’t fit
| Hardware | Q4 need vs. usable | Q4 / Q8 / FP16 | Check price |
|---|---|---|---|
| GeForce RTX 3050 8GB8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 3060 12GB12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 3060 Ti8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 30708 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 3080 10GB10 GB usable | 4.8 / 10.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 3080 Ti12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 309024 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 3090 Ti24 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 40608 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 4060 Ti 16GB16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 407012 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 4070 Super12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 4070 Ti Super16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 4080 Super16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 409024 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 507012 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 508016 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 509032 GB usable | 4.8 / 32.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 7600 XT16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 7800 XT16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 7900 XT20 GB usable | 4.8 / 20.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 7900 XTX24 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Intel Arc B58012 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| Intel Arc Pro B70 32GB32 GB usable | 4.8 / 32.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Mac mini M4 16GB12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| Mac mini M6 16GB12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| Mac mini M6 24GB18 GB usable | 4.8 / 18.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Mac mini M6 32GB24 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Mac mini M5 Pro 64GB48 GB usable | 4.8 / 48.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| MacBook Pro M4 Max 128GB96 GB usable | 4.8 / 96.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| MacBook Pro M5 Max 128GB96 GB usable | 4.8 / 96.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| MacBook Pro M6 32GB24 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Mac Studio M5 Ultra 256GB192 GB usable | 4.8 / 192.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Mac Studio M5 Ultra 512GB384 GB usable | 4.8 / 384.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Framework Desktop (Ryzen AI Max+ 395) 128GB96 GB usable | 4.8 / 96.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA DGX Spark 64GB48 GB usable | 4.8 / 48.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA DGX Spark 128GB96 GB usable | 4.8 / 96.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 2060 6GB6 GB usable | 4.8 / 6.0 GB | Q4FitsQ8Won’t fitFP16Won’t fit | AmazoneBay |
| GeForce RTX 2060 12GB12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 2060 Super8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 20708 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 2070 Super8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 20808 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 2080 Super8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 2080 Ti11 GB usable | 4.8 / 11.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| NVIDIA Titan RTX24 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 3070 Ti8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 3080 12GB12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 4060 Ti 8GB8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 4070 Ti12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| GeForce RTX 408016 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 50608 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 5060 Ti 8GB8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| GeForce RTX 5060 Ti 16GB16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| GeForce RTX 5070 Ti16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 6700 XT12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| Radeon RX 6750 XT12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| Radeon RX 680016 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 6800 XT16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 6900 XT16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 6950 XT16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 76008 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| Radeon RX 7700 XT12 GB usable | 4.8 / 12.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| Radeon RX 7900 GRE16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 9060 XT 8GB8 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| Radeon RX 9060 XT 16GB16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 907016 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Radeon RX 9070 XT16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Intel Arc A7508 GB usable | 4.8 / 8.0 GB | Q4FitsQ8TightFP16Won’t fit | AmazoneBay |
| Intel Arc A770 16GB16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| Intel Arc B57010 GB usable | 4.8 / 10.0 GB | Q4FitsQ8FitsFP16Won’t fit | AmazoneBay |
| NVIDIA RTX A400016 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA RTX A500024 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA RTX A600048 GB usable | 4.8 / 48.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA RTX 2000 Ada 16GB16 GB usable | 4.8 / 16.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA RTX 4000 Ada 20GB20 GB usable | 4.8 / 20.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA RTX 6000 Ada 48GB48 GB usable | 4.8 / 48.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA RTX Pro 6000 Blackwell 96GB96 GB usable | 4.8 / 96.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA Tesla P40 24GB24 GB usable | 4.8 / 24.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| NVIDIA Tesla V100 32GB (PCIe)32 GB usable | 4.8 / 32.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| AMD Radeon Instinct MI50 32GB32 GB usable | 4.8 / 32.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
| AMD Radeon Instinct MI60 32GB32 GB usable | 4.8 / 32.0 GB | Q4FitsQ8FitsFP16Fits | AmazoneBay |
We may earn a commission when you buy through these links, at no extra cost to you.
Quick answers
- How much VRAM does gemma-3n-E2B-it need?
- About 4.8 GB at Q4_K_M, 7.6 GB at Q8_0 and 13.5 GB at FP16. These are estimates: weights plus about 15% for the KV cache at a ~4k context, plus 1 GB for the runtime.
- What is the cheapest GPU that runs gemma-3n-E2B-it?
- Of the GPUs we track, the Intel Arc B570 (10 GB, $219 list price) is the cheapest one that runs it fully on the GPU at Q4.
- Can a Mac or mini PC run gemma-3n-E2B-it?
- Yes. The cheapest Mac or unified-memory mini PC we track that runs it at Q4 is the Mac mini M4 16GB (16 GB, 12.0 GB usable for models, $789).
- How many GPUs and Macs can run gemma-3n-E2B-it?
- 82 of the 82 GPUs and Macs we track run it fully at Q4.