community pool / model answer

Whisper Large v3

audio · openai/whisper-large-v3

4.8 ×real · best measured cell (l4-24gb) · 5 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
L4 24GB (modal)fp164.8 ×realmeasured-4.1 GB-whisper-large-v3

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.