Google’s Gemini 2.5 Flash introduces ‘thinking budgets’ that cut AI costs by 600% when turned down

Google has upgraded its AI offering, giving businesses more control over the ‘thinking’ processed by their AI and enabling developers to tailor computational power depending on the demands of a given task.
Released as Gemini 2.5 Flash in preview through Google AI Studio and Vertex AI, the new model introduces a ‘thinking budget’, allowing developers to specify computational power for complex problems.
Cost adjustments reflect the computational intensity of tasks, with output costs ranging from $0.60 t o$ 3.50 per million tokens, depending on the level of reasoning enabled.
The new pricing model represents a strategic effort to improve reasoning capabilities while maintaining competitive prices, as the AI market becomes increasingly crowded.
Also this week, Google unveiled Veo 2 video generation for Gemini Advanced users and free access to Gemini Advanced for U.S. college students until 2026.

Fast Feed