Google launches Gemini 3.5 Flash, default AI model for agents

Google unveiled Gemini 3.5 Flash at I/O 2026, positioning it as the default model for the Gemini app, AI Mode in Search, and its new personal agent Gemini Spark, while claiming it runs up to four times faster than comparable frontier models and often costs less than half as much [1]. The launch marks a shift from answering questions to executing multi‑step, agentic workflows such as coding, planning, and audit automation, a capability Google says can cut days‑long developer tasks to minutes.

Agentic performance and benchmark scores

Gemini 3.5 Flash outperforms the earlier Gemini 3.1 Pro on coding and agentic benchmarks, posting 76.2 % on Terminal‑bench 2.1, 1,656 Elo on GDPval‑AA, and 83.6 % on MCP Atlas, plus 84.2 % on the multimodal CharXiv Reasoning test [1]. Google ties these results to its Antigravity platform, which can spin up multiple sub‑agents in parallel, enabling more demanding workloads. The model is also marketed as the engine behind Gemini Spark, a 24/7 personal AI assistant currently in a trusted‑tester rollout and slated for a broader beta to Google AI Ultra subscribers in the U.S. next week [1].

Market positioning and enterprise outlook

By making Gemini 3.5 Flash the default, Google is betting that the next AI wave will be judged on actionability rather than pure answer quality. Enterprise customers can access the model via the Gemini Enterprise Agent Platform and Gemini Enterprise, while a Gemini 3.5 Pro version is in internal testing and expected to roll out next month [1]. Analysts note that the model’s enterprise value will hinge on reliability in live workflows, a factor that could differentiate Google from rivals still focused on chatbot‑style interactions [3].

What to watch

• Next‑month rollout of Gemini 3.5 Pro – will extend the agentic capabilities to a higher‑tier offering.
• Enterprise adoption metrics – usage and reliability data from early enterprise pilots will indicate commercial traction.
• Competitive response – how other AI providers adjust pricing or speed claims in light of Google’s “four‑times faster, half‑cost” positioning.

Google’s move pushes generative AI deeper into production environments, making speed and cost the new battleground while testing whether agentic models can reliably replace manual workflows in real‑time business settings.