Google launches Gemini 3.5 Flash for enterprise automation and coding
The tech giant’s latest artificial intelligence release targets complex agentic workflows, offering significantly faster output speeds and lower costs for enterprise partners.
Google has officially released Gemini 3.5 Flash, a new artificial intelligence model designed to handle complex agentic workflows and coding tasks. Announced at the Google I/O developer conference, the model represents a strategic shift towards autonomous agents capable of planning and executing multi-step work with minimal human intervention. It is now generally available globally via the Gemini app, AI Mode in Search, and various developer platforms including Google Antigravity and the Gemini Enterprise Agent Platform.
The model is positioned as a high-speed alternative to flagship offerings, delivering output speeds of approximately 300 tokens per second. Google states this rate is four times faster than other frontier models, while operating at less than half the cost for comparable tasks. In benchmark testing, 3.5 Flash outperformed Gemini 3.1 Pro on specific metrics, including Terminal-Bench 2.1, GDPval-AA, MCP Atlas, and CharXiv Reasoning, positioning it as a strong option for long-horizon tasks that require both intelligence and latency management.
Several enterprise partners are already piloting the model to streamline operations. Macquarie Bank is testing the system to accelerate customer onboarding by reasoning over complex documents exceeding 100 pages. Shopify is utilising subagents for long-horizon data analysis to improve merchant growth forecasts, while Salesforce has integrated the model into its Agentforce platform to automate complicated enterprise tasks.
Other organisations are deploying the technology for specific operational needs. Xero is using agents to manage multi-week workflows such as identifying suppliers and preparing tax forms. Ramp is enhancing optical character recognition for complex invoices, and Databricks is employing agentic workflows for real-time data monitoring and issue diagnosis. These partnerships highlight the model’s capacity to reduce toil in sectors ranging from financial services to data science.
Alongside the model release, Google began rolling out Gemini Spark, a personal AI agent powered by 3.5 Flash, to trusted testers. A beta version is planned for Google AI Ultra subscribers in the US next week. The company also confirmed that Gemini 3.5 Pro is currently in internal use and is scheduled for public release next month. The model was developed under an updated Frontier Safety Framework featuring strengthened cyber safeguards and interpretability tools to monitor the AI’s reasoning processes.
