community pool / model answer

phi-4

chat · microsoft/phi-4

36 tok/s · model median · 3 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
RTX 3060 12GB4-bit36 tok/sreported-8.7 GB2,048ollama
RTX 3080 12GB4-bit77 tok/sreported-8.7 GB2,048llama.cpp
RTX3060+RTX2080-pooled-20GB 20GB4-bit36 tok/sreported-8.7 GB2,048ollama

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.