Google releases Gemini 3.5 Transcribe with 70% faster voice-to-text
Ars Technica AI · rss · 2026-08-27
Google announced Gemini 3.5 Transcribe, a model designed to streamline voice input by editing out disfluencies and corrections. Already powering the Gboard "Rambler" feature on Pixel 11, it will expand across the Google ecosystem. The new model is approximately 70% faster than its predecessor, Chirp 3, reducing the live-speech error rate from 7.32% to 5.5%.
More from Models
- Unredacted Anthropic filing exposes internal data-quality debate months before Opus 3 cutoff — zacharynado · 2026-10-15
- Google says teams across the company are using Gemini 4 Argon on its hardest engineering problems — CurieuxExplorer · 2026-10-15
- Perplexity's new embedding models: index with a 9B model, search with 0.6B — antoine_chaffin · 2026-10-15
- 797GB of n-gram counts in CPU RAM let 1B models match 57% more compute — PangWeiKoh · 2026-10-15
- Leak: Moonshot's Kimi flagship 'x4' done with pre-training, K3.1 in internal testing — lyraxana · 2026-10-15
- Unsloth ships GGUF quants for Nemotron 3.5: Bolt-74B runs in 24GB RAM — danielhanchen · 2026-10-15