Solaria transcription with timestamps, diarization, and translation.
3 integrated models · Official documentation ↗
Released June 2026, Solaria-3 is Gladia's newest model, limited to pre-recorded English, French, German, Spanish and Italian audio, one language per file. Gladia reports 26% lower English WER than Solaria-1 on real customer calls.
Solaria-1, Gladia's April 2025 generalist model, is the default for both pre-recorded and live transcription, covering 100+ languages with code switching. Gladia reports partial transcripts in under 103 ms and finals in about 698 ms.
Gladia's asynchronous transcription API (POST /v2/pre-recorded), which runs Solaria-1 unless another model is named, with Solaria-3 as the opt-in alternative. Requests can add diarization, translation, PII redaction, custom vocabulary and subtitles.