| Model profile | ||
|---|---|---|
| Model | Size | Approximate size |
| Gemma 3 4BModel source: provider131,072 tokens max native context | 4B | 2.4 GB4-bit quantization |
| Compatible hardware Estimated speeds, not benchmark results: calculated from memory bandwidth and model size. Real results can differ significantly because there is no precise formula for deriving LLM generation speed from hardware specifications alone. | ||||
|---|---|---|---|---|
| Hardware | Memory | Bandwidth | vLLM | Estimated generation |
| DDR4 16GB Computer | 16 GBRAM | 51.2 GB/s | no | 11 tok/s4-bit quantization |
| DDR5 16GB Computer | 16 GBRAM | 89.6 GB/s | no | 20 tok/s4-bit quantization |
| DDR5 32GB Computer | 32 GBRAM | 89.6 GB/s | no | 20 tok/s4-bit quantization |
| Mac mini M4 | 24 GBUnified RAM | 120 GB/s | no | 26 tok/s4-bit quantization |
| RTX 2000 Ada | 16 GBGPU VRAM | 224 GB/s | yes | 49 tok/s4-bit quantization |
| Radeon 8060S 96GB | 96 GBUnified RAM | 256 GB/s | yes | 56 tok/s4-bit quantization |
| M4 Pro | 24 GBUnified RAM | 273 GB/s | no | 59 tok/s4-bit quantization |
| DGX Spark | 128 GBUnified RAM | 273 GB/s | yes | 59 tok/s4-bit quantization |
| RTX 4060 Ti 16GB | 16 GBGPU VRAM | 288 GB/s | yes | 62 tok/s4-bit quantization |
| M5 Pro | 64 GBUnified RAM | 307 GB/s | no | 66 tok/s4-bit quantization |
| RTX 3060 12GB | 12 GBGPU VRAM | 360 GB/s | yes | 77 tok/s4-bit quantization |
| RTX 5060 Ti 16GB | 16 GBGPU VRAM | 448 GB/s | yes | 95 tok/s4-bit quantization |
| M5 Max | 128 GBUnified RAM | 614 GB/s | no | 127 tok/s4-bit quantization |
| M3 Ultra | 96 GBUnified RAM | 819 GB/s | no | 164 tok/s4-bit quantization |
| RTX 5080 | 16 GBGPU VRAM | 960 GB/s | yes | 188 tok/s4-bit quantization |
| RTX 3090 Ti | 24 GBGPU VRAM | 1008 GB/s | yes | 196 tok/s4-bit quantization |
| M5 Ultra | 256 GBUnified RAM | 1200 GB/s | no | 227 tok/s4-bit quantization |
| RTX 5090 | 32 GBGPU VRAM | 1792 GB/s | yes | 313 tok/s4-bit quantization |
| RTX PRO 6000 | 96 GBGPU VRAM | 1792 GB/s | yes | 313 tok/s4-bit quantization |
| DGX H200 | 1128 GBGPU HBM3e | 4800 GB/s | yes | 602 tok/s4-bit quantization |
| DGX Station | 748 GBCoherent Memory | 7100 GB/s | yes | 732 tok/s4-bit quantization |
| ET900N G3 | 748 GBCoherent Memory | 7100 GB/s | yes | 732 tok/s4-bit quantization |
| DGX B200 | 1440 GBGPU HBM3e | 8000 GB/s | yes | 771 tok/s4-bit quantization |
| GB200 NVL72 | 13400 GBGPU HBM3e | 8000 GB/s | yes | 771 tok/s4-bit quantization |
| GB300 NVL72 | 20000 GBGPU HBM3e | 8000 GB/s | yes | 771 tok/s4-bit quantization |
Compares Gemma 4B and Gemma 12B outputs on an e-commerce app, connects Claude Desktop via MCP server, and uses DeepEval, Ragas, and Hugging Face Evaluate with LLM-as-a-judge for assessment.