Gemini 2.5: Updates to our family of thinking models- Google Developers Blog

What this post added

Introduces Gemini 2.5 Flash-Lite in preview, offering the lowest latency and cost within the 2.5 model family, optimized for high-throughput tasks like classification and summarization. Updates pricing for Gemini 2.5 Flash, removing the thinking vs. non-thinking price difference and adjusting input/output token costs. Stabilizes Gemini 2.5 Pro (06-05 version) for general availability, noting its continued demand for high-intelligence tasks like coding and agentic applications. Provides transition guidance for users of preview versions of Gemini 2.5 Flash and Pro.