Gemini 3.5 Flash
Best for frontier performance across agents and coding
deepmind.googleHere are the latest publicly announced details on Gemini 3.5 Flash as of May 2026.
Overview
- Gemini 3.5 Flash is Google's speed-focused, agent-capable model released at Google I/O 2026. It emphasizes fast response times, strong coding capabilities, and enhanced agentic features.[2][3]
- It is the default model for the Gemini app and AI Mode in Google Search globally, and is designed to run continuously as your personal AI agent (Gemini Spark) with action-oriented capabilities.[3]
Key capabilities and performance highlights
- Speed and efficiency: Gemini 3.5 Flash delivers much faster output than earlier generations, with claims of significantly higher tokens-per-second throughput and lower latency, enabling smoother multi-step workflows and real-time interactions.[2][3]
- Coding and developer tools: It includes improved coding assistance and is available to developers via Google’s Gemini API, Google AI Studio, Android Studio, and Antigravity (Google’s agent development platform).[6][3]
- Agentic features: The model powers Gemini Spark, a persistent personal AI agent that can operate 24/7, take actions on your behalf under your direction, and support automation across your digital life.[3]
Public availability and roadmap
- Gemini 3.5 Flash is rolling out today as the default model for billions of users in the Gemini app and AI Mode in Google Search, with broader availability across developer and enterprise channels planned or already in progress.[4][3]
- Gemini 3.5 Pro, a larger variant, is in internal use with expectations for a public rollout in the near term (next month or so after the initial 3.5 Flash launch).[4][3]
Context and notable related items
- The Gemini 3 line followed Gemini 3 Pro and 3.1 Pro releases, with 3.5 continuing Google's trend toward frontier intelligence combined with practical action capabilities.[3][4]
- Comparisons in coverage emphasize that 3.5 Flash aims to balance high reasoning and coding prowess with low latency and cost benefits relative to prior flagship models.[2][3]
Illustrative example
- If you ask Gemini Spark to draft a multi-step automation that scrapes a set of pages, summarizes findings, and then generates a report, 3.5 Flash is designed to handle the task with faster iteration cycles and actionable outputs, thanks to its improved agent and coding capabilities.[3]
If you’d like, I can pull out more detailed performance benchmarks Google and third-party coverage cite, or summarize availability by platform (Gemini app, AI Mode in Search, Gemini API, and Enterprise offerings) with direct quotes. I can also help compare Gemini 3.5 Flash to Gemini 3 Pro in a concise, side-by-side table if that would be useful.