Google DeepMind introduced Gemini 3.5, beginning with Flash. The announcement focuses on programming and multi-step workflows with tools, connecting the model to consumer services, development environments, and enterprise applications.
Context
A model and the product around it are different things. The same underlying system can receive different tools, context, and limits in each application. Evaluating a specific assistant therefore requires looking at the whole arrangement, not assuming every capability in the model announcement is exposed everywhere.
Sources & authors
- Gemini 3.5: frontier intelligence with actionGoogle DeepMind · May 19, 2026



