Today marks a significant update to Google’s Gemini 2.5 model family, all of which are “thinking models” capable of reasoning before generating responses, with developers controlling the “thinking budget.” Gemini 2.5 Pro is now generally available and stable, seeing continued high demand for high-intelligence applications like coding and agentic tasks. Gemini 2.5 Flash also reached general availability and stability, with updated pricing: input tokens are now $0.30/1M (up from $0.15), and output tokens are $2.50/1M (down from $3.50), while the distinction between “thinking” and “non-thinking” prices has been removed for simplicity. Additionally, Gemini 2.5 Flash-Lite has been introduced in preview, designed as a cost-effective upgrade from previous Flash models, offering the lowest latency and cost in the 2.5 family for high-throughput tasks such as classification and summarization, with its thinking feature off by default.
https://deepmind.google/discover/blog/gemini-25-updates-to-our-family-of-thinking-models/