Google Debuts Gemini 3.5 Transcribe and 3.7 Flash Models

Google Rolls Out Two Gemini Updates in Quick Succession

Google has pushed out a pair of Gemini model updates aimed at very different use cases: a speech-to-text system that cleans up messy spoken input, and a coding-and-agents model built for developers and automated business workflows. Together, the releases show Google iterating rapidly on its Gemini lineup across both consumer-facing tools and infrastructure aimed at enterprise and developer adoption 123456.

Gemini 3.5 Transcribe Cleans Up How You Talk to Your Phone

The first release, Gemini 3.5 Transcribe, is described by Google as its "most precise speech-to-text model yet" 5. Its standout feature is the ability to strip out verbal filler — the "ums," "ahs," and self-corrections that naturally occur in spoken language — and convert rambling speech into clean, structured text 13. Rather than transcribing speech verbatim, the model interprets intent and tidies up the output so dictated notes, messages, or documents read more like something a person typed carefully rather than spoke off the cuff.

The model is already powering Gboard's Rambler feature, a voice-typing tool on Google's keyboard app, and Google says it will soon extend to any text field within the Chrome browser, effectively turning speech-to-text into a system-wide input option rather than a feature confined to one app 35. Coverage describes this as part of a broader expansion of the technology across Google's ecosystem, suggesting the company sees accurate, clean voice transcription as an increasingly central input method across its products 1.

Gemini 3.7 Flash Targets Coding and AI Agents

Separately, Google released Gemini 3.7 Flash, which it calls its "most intelligent workhorse model yet for coding and agents" 2. The model is positioned for software development tasks and automated business workflows, reflecting the intense competitive pressure among AI developers to win over programmers and enterprises building agentic tools 4.

Notably, Gemini 3.7 Flash arrived just three weeks after the previous 3.6 Flash update, underscoring how fast Google is now shipping incremental model improvements 6. Alongside performance gains, Google cut pricing significantly — reports describe the changes as effectively halving costs — a move likely intended to undercut rivals on affordability for high-volume coding and agent workloads 6.

Reuters' coverage notes that while Google detailed improvements to the Flash tier, it gave no timeline for when its flagship Pro-tier model would arrive, a gap that investors and developers are watching closely given the competitive stakes in the broader AI model race 4.

Why It Matters

The near-simultaneous releases illustrate Google's two-track strategy with Gemini: refining consumer-facing features like transcription that quietly improve everyday products such as Gboard and Chrome, while also racing to keep pace with rivals in coding and agentic capabilities where pricing and performance updates now arrive on a matter of weeks rather than months 123456.