Google launches Gemini 3.5 Flash for agentic AI
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Google launched Gemini 3.5 Flash at I/O 2026 as its "strongest agentic and coding model yet" for long-horizon workflows, making it available in the Gemini app and Search's AI Mode.
- Gemini 3.5 Flash beats Google's own 3.1 Pro on Terminal-Bench 2.1 (76.2%), GDPval-AA (1656 Elo), and MCP Atlas (83.6%), and runs 4x faster than other frontier models — 12x faster in Antigravity — per Koray Kavukcuoglu.
- Artificial Analysis called Gemini 3.5 Flash the clear leader on the Intelligence vs Speed Pareto frontier but flagged it as 5x the cost of Gemini 3 Flash.
- Box AI testing showed a 12 percentage point jump on complex document tasks versus the previous Flash model, per Aaron Levie.
- Google, cited by VentureBeat, claims Gemini 3.5 Flash can cut enterprise AI costs by more than $1 billion a year.
- The Decoder framed the launch as following Anthropic and OpenAI in making newer AI models "significantly pricier."
- GitHub confirmed Gemini 3.5 Flash is "now generally available and rolling out in GitHub Copilot," citing strong tool use and high cache efficiency.
Why it matters: Google is selling Gemini 3.5 Flash as an enterprise cost-cutter (VentureBeat cites $1B+ in claimed annual savings) on the same day Artificial Analysis pegs it at 5x the price of its predecessor — enterprise buyers now face a vendor-versus-analyst split on the very cost story the launch is built around.
Ask SkimNews
