Google announced Gemini 3.8 Flash, rolling out in Antigravity and via the API in Google AI Studio. Alongside the model, it launched the Fairwind Program, giving governments and trusted partners access to a 3.8 Flash Cyber variant for securing vital infrastructure and national security.
A software engineer's one-month review of Google's Antigravity 2.0 with Gemini Flash 3.7 and 3.8 finds 3.7 roughly 1.5-2x faster while 3.8 produces higher-quality output. The reviewer concludes Google has entered agentic coding workflows but still trails Anthropic and OpenAI on features, while offering the most generous plan limits.
A software engineer's one-month review of Google's Antigravity 2.0 with Gemini Flash 3.7 and 3.8 finds 3.7 roughly 1.5-2x faster while 3.8 produces higher-quality output. The reviewer concludes Google has entered agentic coding workflows but still trails Anthropic and OpenAI on features, while offering the most generous plan limits.
Google released Gemini 3.8 Flash, publishing a model card on DeepMind's site alongside a companion Flash Cyber variant. The launch drew heavy discussion on Hacker News and coverage from The Register, and the model reached GitHub Copilot the next day.
Google released Gemini 3.8 Flash, publishing a model card on DeepMind's site alongside a companion Flash Cyber variant. The launch drew heavy discussion on Hacker News and coverage from The Register, and the model reached GitHub Copilot the next day.
A user testing both models reports Gemini 3.8 Flash feels faster, hallucinates less, and gives better answers than 3.1 Pro in the AI Studio web UI, asking whether others see the same. The thread is community impressions rather than an official release announcement.
A user testing both models reports Gemini 3.8 Flash feels faster, hallucinates less, and gives better answers than 3.1 Pro in the AI Studio web UI, asking whether others see the same. The thread is community impressions rather than an official release announcement.