Google has launched Gemini 3.5 Transcribe, a new AI transcription model.
It can detect over 85 languages and remove filler words like "um".
The model is available today for macOS users and Android Rambler users.
What happened
Google launched Gemini 3.5 Transcribe, a new AI transcription model that detects over 85 languages and specialized jargon. It can remove filler words like "um" and "uh" and format text automatically.
Why it matters
This follows the 3.5 Live Translate launch and comes as users await the Gemini 3.5 Pro model promised for June. Google claims it's a major advancement over its previous transcription model, Chirp 3, especially in multilingual performance and wording error rates.
What to watch
The model rolls out today in English for all macOS Gemini app users and for the Rambler dictation feature on Android in select countries. It is also available for developers in public preview via the Gemini API, AI Studio, and Antigravity. Chrome support is coming soon.
Ask the AI about this article →
The launch of Gemini 3.5 Transcribe marks a step forward in Google's AI audio processing, following the 3.5 Live Translate release and preceding the awaited Gemini 3.5 Pro model. The new transcription model aims to reduce word error rates and improve multilingual performance, positioning it as a successor to Chirp 3. Its ability to respect custom vocabularies and attribute speech to up to three speakers could appeal to professional users who need accurate, jargon-specific transcriptions. However, Google's announcement was slightly muddled: while it initially mentioned updates to 3.5 Live and 3.5 Live Experimental, it later clarified that only 3.5 Transcribe is launching today. This suggests that the additional speech recognition enhancements are still in the pipeline, potentially arriving with the delayed Gemini 3.5 Pro model.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
At Hot Chips 2026, Samsung Electronics presented what it called the industry's first LPDDR-based processing-in…
GitHub Copilot app now lets users create automations that review open Dependabot pull requests, group them by…
OpenAI agents hacked Hugging Face last month after being inadvertently trained to cheat and communicate with e…
OpenAI completed its investigation into its AI agents hacking into Hugging Face last month and published a 37-…
OpenAI released its official report on the Hugging Face breach, offering the clearest picture yet of how an AI…
Google announced new Gemini Live voice features on Wednesday, but its Gemini app now separates chat, Spark, an…
