Google Launches Gemini 3.5 Flash at I/O 2026: A Faster and Cheaper Model for AI Agents and Coding

Google recently released Gemini 3.5 Flash at Google I/O May, 2026. The first version of Gemini 3.5. The series combines borderline intelligence and action. Google is calling it a giant leap for intelligent agents. The Flash tier has historically been faster and cheaper. 3.5 Flash outperforms Gemini 3.1 Pro in challenging benchmarks. The previous premium category is being superseded.

What the Ratings Mean

Gemini 3.5 Flash scores 76.2% in Terminal-Bench 2.1. That benchmark tests the performance of the code. It scores 1656 Elo in GDPval-AA. That measures the performance of a real-world agent. It scores 83.6% in the MCP Atlas. The MCP Atlas measures the reliability of standardized instrument use. It scores 84.2% in CharXiv Reasoning. That benchmark tests multimodal understanding.

Gemini 3.5 Flash is 4x faster for output tokens. Jobs are often completed at less than half the cost. The official price is $1.50 for one million input tokens. The output tokens cost $9.00 per million. The saved input is valued at $0.15 per million.

The context window is 1,048,576 input tokens. The maximum output is 65,536 tokens. Input based on text, image, audio, and video. Expiry date is January 2026. Dynamic thinking is on by default. The model automatically allocates more computing power to harder problems.

Designed for Agentic and Long-Horizon operations

Here 'Agentic' refers to programming models, call tools, and replication. It completes multi-step goals, not single questions. 'Long-horizon' means that the loop runs for a long time. Google introduced Managed Agents to the Gemini API. A single API call activates a full agent. It reasons, implements tools, and generates code. The environment runs inside a standalone Linux container. Files and status persist across all follow-up calls. This enables seamless multi-turn agent sessions.

Previously, managing agent status and locations was done manually. The Managed Services API bypasses that infrastructure entirely.

The Antigravity Ecosystem

Google Antigravity is its first agent development platform. Need ideas for production-ready applications. Antigravity 2.0 is a new standalone desktop application. Schedules multiple agents working in parallel. Dynamic subagents manage parallel workflows. Scheduled tasks enable background automation. Integration includes Google AI Studio, Android, and Firebase.

Antigravity CLI is for backend developers. Creates agents instantly, without a GUI. Google encourages Gemini CLI users to migrate now. The Antigravity SDK provides programmed access to the harness. You can define custom agent behavior with it. Agents host the infrastructure of your choice.

Real-World Enterprise Deployments

According to Google, many of our business partners are already using Flash 3.5. Shopify uses subagents in parallel to analyze data. It enables the most accurate global merchant growth forecasts. Macquarie Bank is testing customer onboarding. The model results in complex documents of 100+ pages. Obtains information and makes reliable recommendations.

Salesforce integrates 3.5 Flash into Agentforce. It automates business operations using multiple subagents. Subagents keep context for all complex, multi-tool calls. Ramp uses smart OCR for invoices. It combines multimodal understanding with historical pattern thinking. Xero uses agents for complex, multi-week workflows. One example is collecting supplier data on 1099 forms. Databricks uses an agent workflow for real-time data monitoring. The model diagnoses problems and suggests fixes to developers.

Check it out Technical details. Also, feel free to follow us Twitter and don't forget to join our 150k+ ML SubReddit and Subscribe to Our newspaper. Wait! are you on telegram? now you can join us on telegram too.

Need to work with us on developing your GitHub Repo OR Hug Face Page OR Product Release OR Webinar etc.? contact us

Michal Sutter is a data science expert with a Master of Science in Data Science from the University of Padova. With a strong foundation in statistical analysis, machine learning, and data engineering, Michal excels at turning complex data sets into actionable insights.