FrontierAugust 28, 2026via MarkTechPost
Google AI Releases Gemini 3.5 Transcribe: A Speech-to-Text Model Reporting 2.6% Average WER Across 85+ Languages
Why it matters
Google ships a major speech-to-text capability upgrade with measurable benchmarks (WER, latency, cost) and a deliberate architectural choice (dual endpoints) that practitioners building voice agents need to evaluate against existing transcription stacks.
Key signals
- Gemini 3.5 Transcribe released as two endpoints: streaming (sub-second, 4.0% WER, no diarization/timestamps) and batch (2.6% WER, includes diarization/timestamps, 50% lower cost)
- 85+ language support
- 70% faster finalization vs. Chirp 3
- Streaming vs. batch tradeoff: speed and simplicity vs. feature richness and cost
- Direct competitor to existing transcription services (Whisper, Azure Speech, AWS Transcribe)
The hook
2.6% WER across 85+ languages. Google's Gemini 3.5 Transcribe splits streaming and batch to beat Chirp 3 on speed and cost.
Google has released Gemini 3.5 Transcribe, a speech-to-text model that ships as two separate endpoints rather than one. The streaming endpoint delivers sub-second transcription but drops speaker diarization and word timestamps. The batch endpoint keeps both, at half the cost. Google reports 4.0% wor…