My take wasn't that they are claiming SOTA, just that they are claiming same/better accuracy than whisper "tiny" and "base" models at higher performance versus original whisper.

The WER scores on the larger whisper models are lower (better) than "tiny"/"base", so none of these small models are close to SOTA either way.

Groq and fal are hosters of third party models. They cannot be SOTA.

Fal provides Whisper which is OpenAI.

Groq only hosts LLMs.

Groq and Fal.ai just companies that host, they dont make models