community pool / model answer

Phi-4-mini-reasoning

chat · microsoft/Phi-4-mini-reasoning

97.4 tok/s · model median · 10 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
Radeon 890M 32GB4-bit33.2 tok/sreported-- GB4,096llama.cpp
RTX 3090 24GB4-bit110.6 tok/smeasured1,4030 GB131,072llama.cpp

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.