Google Introduces Gemini 3.5 Flash at I/O 2026: A Faster and Cheaper Model for AI Agents and Coding
Google simply launched Gemini 3.5 Flash at Google I/O May, 2026. It is the primary Gemini 3.5 mannequin. The collection combines frontier intelligence with motion. Google calls it a serious leap for clever brokers. The Flash tier has traditionally been sooner and cheaper. 3.5 Flash outperforms Gemini 3.1 Pro on difficult benchmarks. The earlier premium tier has now been surpassed.
What the Benchmarks Say
Gemini 3.5 Flash scores 76.2% on Terminal-Bench 2.1. That benchmark exams coding efficiency. It scores 1656 Elo on GDPval-AA. That measures real-world agentic process efficiency. It scores 83.6% on MCP Atlas. MCP Atlas measures scaled tool-use reliability. It scores 84.2% on CharXiv Reasoning. That benchmark exams multimodal understanding.
Gemini 3.5 Flash is 4x sooner on output tokens. Tasks usually full at lower than half the fee. Official pricing is $1.50 per million enter tokens. Output tokens price $9.00 per million. Cached enter is priced at $0.15 per million.
The context window is 1,048,576 enter tokens. Maximum output is 65,536 tokens. Supported inputs are textual content, picture, audio, and video. The information cutoff is January 2026. Dynamic pondering is on by default. The mannequin auto-allocates extra compute for more durable issues.
Built for Agentic and Long-Horizon Tasks
Here ‘Agentic’ means the mannequin plans, calls instruments, and iterates. It completes multi-step targets, not single questions. ‘Long-horizon’ signifies that loop runs for prolonged intervals. Google launched Managed Agents within the Gemini API. One API name spins up a full agent. It causes, makes use of instruments, and executes code. The atmosphere runs inside an remoted Linux container. Files and state persist throughout follow-up calls. This allows seamless multi-turn agent periods.
Previously, managing agent state and environments was guide. The Managed Agents API abstracts that infrastructure totally.
The Antigravity Ecosystem
Google Antigravity is its agent-first growth platform. It takes concepts to production-ready apps. Antigravity 2.0 is a brand new standalone desktop app. It orchestrates a number of brokers working in parallel. Dynamic subagents deal with parallelized workflows. Scheduled duties allow background automation. Integrations cowl Google AI Studio, Android, and Firebase.
The Antigravity CLI is for terminal-based builders. It creates brokers immediately, with no GUI. Google encourages Gemini CLI customers emigrate now. The Antigravity SDK provides programmatic entry to the harness. You can outline customized agent behaviors with it. Host brokers on infrastructure of your alternative.
Real-World Enterprise Deployments
According to Google, a number of enterprise companions are already working 3.5 Flash. Shopify runs subagents in parallel for knowledge evaluation. It powers extra correct service provider development forecasts globally. Macquarie Bank is piloting it for buyer onboarding. The mannequin causes over advanced 100+ web page paperwork. It retrieves info and makes dependable suggestions.
Salesforce is integrating 3.5 Flash into Agentforce. It automates enterprise duties utilizing a number of subagents. Subagents retain context throughout advanced, multi-turn software calling. Ramp makes use of it for smarter OCR on invoices. It pairs multimodal understanding with historic sample reasoning. Xero deploys brokers for advanced, multi-week workflows. One instance is gathering provider knowledge for 1099 varieties. Databricks makes use of agentic workflows for real-time knowledge monitoring. The mannequin diagnoses points and proposes fixes for engineers.
Check out the Technical details. Also, be happy to observe us on Twitter and don’t overlook to affix our 150k+ ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Need to associate with us for selling your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar and many others.? Connect with us
The publish Google Introduces Gemini 3.5 Flash at I/O 2026: A Faster and Cheaper Model for AI Agents and Coding appeared first on MarkTechPost.
