Google has unveiled Gemini 3.5, a new family of artificial intelligence models focused on coding, reasoning, and long-horizon agentic workflows. The rollout begins today with Gemini 3.5 Flash, a model designed to execute complex multi-step tasks with improved speed and efficiency.
Google says 3.5 Flash outperforms its previous Gemini 3.1 Pro model on several coding and agentic benchmarks, including Terminal-Bench 2.1, MCP Atlas, and CharXiv Reasoning. The company also claims the model generates output tokens four times faster than other frontier models. According to Google, this enables faster execution of tasks such as maintaining codebases, analyzing large datasets, and automating complex workflows.
Alongside the model upgrade, Google introduced Gemini Spark, a personal AI agent powered by 3.5 Flash. Google says the tool can operate continuously and take action on a user's behalf under supervision. The announcement reflects a growing industry push toward persistent AI assistants, as we've seen with Perplexity's Personal Computer and OpenAI's Codex app.
Google says Gemini 3.5 was developed with expanded safeguards and new safety training designed to reduce harmful outputs while lowering the rate of incorrectly refused safe requests.
Gemini 3.5 Flash is available now globally through the Gemini app, AI Mode in Search, Google AI Studio, Android Studio, and Google's enterprise AI platforms. Gemini Spark is rolling out first to trusted testers, with a beta planned for Google AI Ultra subscribers in the United States next week. Google says Gemini 3.5 Pro is currently in internal testing and scheduled for release next month.