Google just released the Gemini 3.5 Flash at Google I/O in May 2026. It is the first Gemini 3.5 model. The series combines borderline intelligence with action. Google calls it a big leap for savvy customers. Flash has historically been faster and cheaper. 3.5 The Flash beats the Gemini 3.1 Pro in hard benchmarks. The previous premium class has now been transcended.
What do the standards say?
Gemini 3.5 Flash scored 76.2% on Terminal-Bench 2.1. This benchmark tests encoding performance. It has a 1656 Elo on GDPR Val-AA. This measures the agent’s task performance in the real world. It received a score of 83.6% on the MCP Atlas. The MCP Atlas measures the reliability of using graded instruments. It scored 84.2% on CharXiv reasoning. This standard tests understanding of multimedia.
Gemini 3.5 Flash is 4x faster at outputting codes. Tasks are often completed for less than half the cost. The official price is $1.50 per million input tokens. Output tokens cost $9.00 per million. The price for cached inputs is $0.15 per million.
The context window is 1,048,576 input characters. Maximum output is 65,536 tokens. Supported inputs are text, image, audio and video. The deadline for knowledge is January 2026. Dynamic thinking is on by default. The model automatically allocates more computation to more difficult problems.
Designed for proxy and long-term missions
Here “Agentic” means typical plans, callbacks and iterations. It completes multi-step objectives, not single questions. “Long horizon” means that the loop runs for long periods. Google has introduced managed agents in the Gemini API. A single API call creates a complete proxy. It thinks, uses tools, and executes code. The environment runs inside an isolated Linux container. Files and status persist via follow-up calls. This enables seamless multi-turn agent sessions.
Previously, agent state and environments were managed manually. The Managed Agents API encapsulates that entire infrastructure.
Anti-gravity ecosystem
Google Antigravity is the premier agent development platform. Takes ideas into production-ready applications. Antigravity 2.0 is a new standalone desktop application. It coordinates multiple agents working in parallel. Dynamic subagents handle parallel workflows. Scheduled tasks allow for background automation. Integrations cover Google AI Studio, Android, and Firebase.
Antigravity CLI is intended for terminal-based developers. Creates agents instantly, without a GUI. Google encourages Gemini CLI users to migrate now. The Antigravity SDK provides programmatic access to the belt. You can define custom agent behaviors using it. Host proxies on the infrastructure of your choice.
Real-world enterprise deployments
According to Google, several enterprise partners are already running Flash 3.5. Shopify runs sub-agents in parallel to analyze data. It supports more accurate forecasts of merchant growth globally. Macquarie Bank is piloting it for customer onboarding. The form grounds over 100+ pages of complex documents. It retrieves information and makes reliable recommendations.
Salesforce is integrating Flash 3.5 into Agentforce. It automates enterprise tasks using multiple sub-agents. Sub-agents maintain context by invoking complex, multi-role artifacts. Ramp uses it for smarter OCR on invoices. It combines multimodal understanding with historical typological thinking. Xero deploys agents for complex, multi-week workflows. One example is collecting supplier data for 1099 forms. Databricks uses proxy workflows to monitor data in real time. The model diagnoses problems and suggests fixes to engineers.
verify Technical details. Also, feel free to follow us on twitter Don’t forget to join us 150k+ mil SubReddit And subscribe to Our newsletter. I am waiting! Are you on telegram? Now you can join us on Telegram too.
Do you need to partner with us to promote your GitHub Repo page, face hug page, product release, webinar, etc.? Contact us
