Released **MAI-Transcribe-2**, a speech-recognition model reported to rank first on the FLEURS benchmark across 60 languages.

Microsoft released MAI‑Transcribe‑2 in public preview through Microsoft Foundry/Azure Speech. It claims No. 1 on FLEURS across 60 languages, at 5.2% average word-error rate. The model adds speaker diarization and word-level timestamps—useful for meetings, captions and call centers—and is priced at $0.10 per audio hour through 2026.