Understand this story
On May 19, 2026, Google DeepMind released the Gemini 3.5 series, debuting the Gemini 3.5 Flash model first. According to its introduction, the model scored 76.2% on Terminal-Bench 2.1, outputs at 4 times the speed of other frontier models, has been fully opened up, and has become the default model for the Gemini app and AI Mode in global Search; Gemini 3.5 Pro is planned for launch next month. On May 20, Google's Noam Shazeer said that Gemini 3.5's combination of performance and speed makes it suitable for long-horizon and agentic tasks. On the same day, separate reports said that Gemini Spark, a 24/7 personal AI agent powered by Gemini 3.5, can work across apps such as Docs, Calendar, and Slides, and keeps running after the computer is turned off, with rollout starting next week.
AI-generated from public reports
Latest development · Report
English translation is being prepared.Developments
2 confirmed developments
- Source:Ammaar Reshi
- Source:Google DeepMind
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
May 21
May 20
Noam Shazeer
Noam Shazeer says Gemini 3.5 is suited for long-horizon and agentic tasks
Google's Noam Shazeer said that Gemini 3.5's combination of performance and speed makes it suited for long-horizon and agentic tasks, and that enterprise partners have already seen real results in their daily workflows and accelerated their workflows. He attached a blog link introducing Gemini 3.5.
noamshazeer
English translation is being prepared.
May 19
google_deepmindFirst-partySelected
English translation is being prepared.