Loading…
Loading…
Models / gemma-4-E2B / gemma-4-E2B ~5.12B
Any to any · served with vLLM · ↓ 44.4K downloads a month · View on Hugging Face
cyankiwi/gemma-4-E2B-it-AWQ-INT4 needs about 11 GB of GPU memory for conversations of up to 8,192 tokens. It runs on one RTX 3060 from $0.09/h right now; best value: RTX PRO 6000 at $1.13/h.
The same model packaged differently: a quantized version needs less GPU memory and runs on a cheaper machine.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalgoogle/gemma-4-E2B | vLLM | 10 GB | 13.9 GB | $0.13/h RTX 4060 Ti |
| itgoogle/gemma-4-E2B-it | vLLM | 10 GB | 13.9 GB | $0.13/h RTX 4060 Ti |
| it qat w4a16 ctgoogle/gemma-4-E2B-it-qat-w4a16-ct | vLLM | 8.3 GB | 11.8 GB | $0.09/h RTX 3060 |
| it qat q4 0 unquantizedgoogle/gemma-4-E2B-it-qat-q4_0-unquantized | vLLM | 10 GB | 13.9 GB | $0.13/h RTX 4060 Ti |
| it qat mobile ctgoogle/gemma-4-E2B-it-qat-mobile-ct | vLLM | 2.7 GB | 5.5 GB | $0.09/h RTX 3060 |
| it unsloth bnb 4bit (unsloth)unsloth/gemma-4-E2B-it-unsloth-bnb-4bit | vLLM | 8.1 GB | 11.6 GB | $0.09/h RTX 3060 |
$1.13/h from your balance while it runs, billed by the hour; stop it from its page. GridPup rents the machine, which downloads the model from Hugging Face (usually ready in 10 to 15 minutes), and gives you an https address, keys and a Playground.
| it AWQ INT4 (cyankiwi)cyankiwi/gemma-4-E2B-it-AWQ-INT4 | vLLM | 7.6 GB | 11 GB | $0.09/h RTX 3060 |
| gemma 4 E4B it ONNX (onnx-community)onnx-community/gemma-4-E4B-it-ONNX | ONNX Runtime | 100 MB | 1.2 GB | $0.07/h Quadro P4000 |
| it ONNX (onnx-community)onnx-community/gemma-4-E2B-it-ONNX | ONNX Runtime | 100 MB | 1.1 GB | $0.07/h Quadro P4000 |
| it qat mobile ONNX (onnx-community)onnx-community/gemma-4-E2B-it-qat-mobile-ONNX | ONNX Runtime | 90 MB | 1.1 GB | $0.07/h Quadro P4000 |
Prices are per hour for the whole machine and its disk, right now. Speed is Vast's DLPerf score of the machine, relative to the cheapest. Best value: the most speed per dollar. Only machines with a Vast reliability above 95% that download at 400 Mbit/s or more are offered: the machine fetches the model itself.