community pool / model answer
gemma-4-12B-it-qat-q4_0-gguf
chat · google/gemma-4-12B-it-qat-q4_0-gguf
18.9 tok/s · model median · 8 runs in the model record.
| Reference rig | Quantization | Median result | Basis | TTFT (ms) | Peak VRAM | Max context | Engines |
|---|---|---|---|---|---|---|---|
| RTX 3080 12GB | 4-bit | 71.2 tok/s | reported | 968 | 8.6 GB | 262,144 | llama.cpp |
| RTX 5090 32GB | 4-bit | 130.8 tok/s | reported | 100 | - GB | 2,048 | lmstudio |
| RX 570 4GB ×2 | 4-bit | 16.7 tok/s | measured | 737 | - GB | 2,048 | llama.cpp |
| RX 7700 XT 12GB | 4-bit | 40.4 tok/s | reported | 226 | - GB | 2,048 | vllm |
Context tested
The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.