The model beats its previous version on all the benchmarks published by Google, and outperforms Claude Opus 5 on three of them.
Google released on September 2 Gemini 3.8 Flash, with a locked release called 3.8 Cyber Flash for cybersecurity tasks. The model processes text, image, audio, video and PDF with a context window of 1 million tokens and up to 64,000 tokens output, designed for code over long horizons and autonomous agents. Built on the basis of 3.7 Flash rather than a new model, it "works more" by consuming more tokens of reflection — Google recommends staying on 3.7 Flash for pure efficiency uses.
The rate remains at $0.75 per million tokens on entry and $3.75 on exit, but only until December 31, 2026: prices will double on January 1, 2027. The model is already available in the Gemini app for Google AI Pro and Ultra subscribers, in AI Mode, in Google Sheets, as well as for developers via Google Antigravity, AI Studio and the Gemini API.
Read the full article on the original source
Go to the source →