Codesota · GuidesEditorial deep-dives on models, benchmarks & evaluation28 guides · 7 sections
§ Guides

Long-form, dated.

The guides archive. Each piece takes a model family, a benchmark, or a method and treats it as a subject in its own right — with the same standard of evidence we apply to the registry itself. Grouped by what the guide is about, not by when it was written.

§ 03

Speech & audio

ASR and TTS compared on the numbers that matter: WER, MOS, latency and price per hour.

1 guide
16
Speech recognition: Whisper vs Gemini vs Deepgram
WER on LibriSpeech, pricing per hour, latency, streaming support. An ASR model showdown.
§ Elsewhere

Read sideways.

News
Dated dispatches on model releases and benchmark shifts.
Methodology
How scores are verified and how retractions are recorded.
Benchmarks
The live registry. Every task, every score, every date.
Papers with Code
What happened to the archived Meta registry — and the replacement.