Loading…
Loading…
Models / Llama-3.2
4 models (1B to 3B) · ↓ 8.2M downloads a month · from $0.08/h on GTX 1660 SUPER right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalmeta-llama/Llama-3.2-1B-InstructLicence to accept | vLLM | 2.5 GB | ~5.2 GB | $0.10/h RTX 3060 |
| FP8 dynamic (RedHatAI)RedHatAI/Llama-3.2-1B-Instruct-FP8-dynamic | vLLM | 2.0 GB | 4.8 GB | $0.10/h RTX 3060 |
| FP8 (RedHatAI)RedHatAI/Llama-3.2-1B-Instruct-FP8 | vLLM | 2.0 GB | 4.8 GB | $0.10/h RTX 3060 |
| unsloth bnb 4bit (unsloth)unsloth/Llama-3.2-1B-Instruct-unsloth-bnb-4bit | vLLM | 1.1 GB | 3.7 GB | $0.10/h RTX 3060 |
| bnb 4bit (unsloth)unsloth/Llama-3.2-1B-Instruct-bnb-4bit | vLLM | 1.0 GB | 3.7 GB | $0.10/h RTX 3060 |
| quantized.w8a8 (RedHatAI)RedHatAI/Llama-3.2-1B-Instruct-quantized.w8a8 | vLLM | 2.0 GB | 4.8 GB | $0.10/h RTX 3060 |
| llama 3 2 1b instruct onnx genai int4 kquantlast emb int8 vivo x300 (Arm)Arm/llama-3-2-1b-instruct-onnx-genai-int4-kquantlast-emb-int8-vivo-x300 | ONNX Runtime GenAI | 1.2 GB | 3.1 GB | $0.08/h GTX 1660 SUPER |
| llama 3 2 1b instruct onnx genai int4 kquantlast emb int8 graviton g4 (Arm)Arm/llama-3-2-1b-instruct-onnx-genai-int4-kquantlast-emb-int8-graviton-g4 | ONNX Runtime GenAI | 1.2 GB | 3.2 GB | $0.08/h GTX 1660 SUPER |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalmeta-llama/Llama-3.2-1BLicence to accept | vLLM | 2.5 GB | ~5.2 GB | $0.10/h RTX 3060 |
| llama 3 2 1b base onnx genai int4 kquantlast emb int8 vivo x300 (Arm)Arm/llama-3-2-1b-base-onnx-genai-int4-kquantlast-emb-int8-vivo-x300 | ONNX Runtime GenAI | 1.2 GB | 3.1 GB | $0.08/h GTX 1660 SUPER |
| llama 3 2 1b base onnx genai int4 kquantlast emb int8 graviton g4 (Arm)Arm/llama-3-2-1b-base-onnx-genai-int4-kquantlast-emb-int8-graviton-g4 | ONNX Runtime GenAI | 1.2 GB | 3.2 GB | $0.08/h GTX 1660 SUPER |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalmeta-llama/Llama-3.2-3B-InstructLicence to accept | vLLM | 6.4 GB | ~9.9 GB | $0.10/h RTX 3060 |
| FP8 dynamic (RedHatAI)RedHatAI/Llama-3.2-3B-Instruct-FP8-dynamic | vLLM | 4.4 GB | 8.2 GB | $0.10/h RTX 3060 |
| FP8 (RedHatAI)RedHatAI/Llama-3.2-3B-Instruct-FP8 | vLLM | 4.4 GB | 8.1 GB | $0.10/h RTX 3060 |
| unsloth bnb 4bit (unsloth)unsloth/Llama-3.2-3B-Instruct-unsloth-bnb-4bit | vLLM | 2.4 GB | 5.9 GB | $0.10/h RTX 3060 |
| bnb 4bit (unsloth)unsloth/Llama-3.2-3B-Instruct-bnb-4bit | vLLM | 2.2 GB | 5.8 GB | $0.10/h RTX 3060 |
| llama 3 2 3b instruct onnx genai int4 kquantlast emb int8 (Arm)Arm/llama-3-2-3b-instruct-onnx-genai-int4-kquantlast-emb-int8 | ONNX Runtime GenAI | 2.7 GB | 5.6 GB | $0.08/h GTX 1660 SUPER |
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| Originalmeta-llama/Llama-3.2-3BLicence to accept | vLLM | 6.4 GB | ~9.9 GB | $0.10/h RTX 3060 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").