Model registry: Large v3 speed label corrected to measured ~0.3-0.4x real-time (CPU int8 quiet-machine benchmark Sep 22: 0.27x on 300s, 0.43x on 1200s; prior '~1x' was optimistic)

2026-09-22 10:38:34 -05:00 · 2026-09-22 10:38:34 -05:00 · b4e335d720
commit b4e335d720
parent 89a882cb55
1 changed files with 1 additions and 1 deletions
--- a/backend/shonar/services/ai/model_registry.py
+++ b/backend/shonar/services/ai/model_registry.py
@ -53,7 +53,7 @@ SUPPORTED_TRANSCRIPTION_MODELS: dict[str, TranscriptionModelInfo] = {
        "~3 GB once; slow on CPU — best on machines with 16 GB+ RAM.",
        params="~1.5B",
        approx_memory="~10 GB RAM",
-        relative_speed="~1x real-time (CPU)",
+        relative_speed="~0.3–0.4x real-time (CPU)",
    ),
 }