Google introduced Gemini 3.5, its latest family of artificial intelligence models that combine frontier capabilities with autonomous action. The launch marks a significant advancement in building more capable intelligent agents, with the Gemini 3.5 Flash available globally starting today. The model is purpose-built to run complex, agentic workflows at scale, delivering cutting-edge performance at exceptional speeds.
The Gemini 3.5 Flash delivers intelligence rivaling large flagship models in multiple dimensions while maintaining the speeds expected from the Flash series. Supera o Gemini 3.1 Pro in challenging coding benchmarks and agentic tasks such as Terminal-Bench 2.1 with 76.2%, GDPval-AA with 1656 Elo and MCP Atlas with 83.6%. In multimodal comprehension, it achieves 84.2% on CharXiv Reasoning. The model processes output tokens four times faster than other frontier models.
Disponibilidade global and developer access
The Gemini 3.5 Flash is accessible globally through multiple channels. Google offers the model via Gemini app and Mode AI on Busca Google for common users. Desenvolvedores can access the technology through the Google Antigravity agent-first development platform, Gemini API on Google AI Studio and Android Studio. Empresas has access via Gemini Enterprise Agent Platform and Gemini Enterprise.
Google is also working on the development of Gemini 3.5 Pro. Já is being used internally and the company plans to launch it publicly next month. The model follows the same principle of combining frontier intelligence with autonomous action, expanding agentic capabilities for more demanding use cases.
Desempenho in complex agentic tasks
The balance between speed and performance makes the Gemini 3.5 Flash ideal for handling long-horizon agentic tasks. What took developers days to complete, or auditors weeks to complete, the model can now help complete in a fraction of the time, often at a cost of less than half that of other frontier models. The system quickly plans, builds and iterates to solve real-world problems, whether developing new applications, maintaining code bases or preparing financial documents. Acoplado to the updated Antigravity harness, the model becomes a powerful engine for implementing collaborative subagents that tackle problems at scale for more demanding use cases.
Sob supervision, the Gemini 3.5 Flash performs multi-step workflows and encoding tasks while maintaining frontier performance. Google demonstrated the model’s ability to automate renaming and categorization of unstructured assets based on dynamic criteria. Parceiros development teams are already reporting significant progress. Bancos and fintechs automate multi-week workflows. Data Science Equipes extract insights from complex data environments.
Capacidades multimodal and graphical interfaces
Construído builds on the strong multimodal foundation of Gemini 3, the new model generates richer, more interactive web UIs and graphics. Demonstrações internals showcase Gemini 3.5 Flash creating interactive animations for academic research through AI Studio. The ability to generate dynamic visual interfaces expands practical applications in software development, data visualization, and personalized user experiences.
Real Impacto with enterprise use cases
Gemini 3.5 Flash is already driving significant progress for developers and businesses. Durante 3.5 series development, Google worked with industry partners to understand where toil and complexity arise in real workflows:
- Shopify runs subagents in parallel to analyze complex long-horizon data, generating more accurate predictions of merchant growth on a global scale
- Instituições financial institutions automate processes that previously took weeks
- Data Science Equipes Identify Hidden Patterns in Unstructured Data Environments
- Empresas technology reduces new product development time
Gemini Spark and personal AI agents
Gemini 3.5 Flash is now the default template for Gemini and AI Mode application in Busca Google globally. Google introduced Gemini Spark, new personal AI agent powered by Flash 3.5. Opera 24/7 helping users navigate digital life by taking action on your behalf while under your direction. The company began rollout to trusted testers today, with plans to bring the Beta to Google AI Ultra subscribers in the US next week.
Flash 3.5’s enhanced agentic coding capabilities also deliver smarter experiences on Busca. Novos information agents work 24/7. Dynamic Generative UI Experiências has been unlocked. Busca leverages the model to build interactive visualizations explaining complex patterns.
Integrated Frontier Segurança
Gemini 3.5 is developed in accordance with Frontier Safety Framework of Google. The company has strengthened cyber and CBRN protections, making it less likely to generate harmful content and reducing mistaken denials of secure queries. Isso is achieved with advanced security training and mitigations, including interpretability tools that verify and understand the AI’s internal reasoning before providing response.
