Google Launches Advanced Gemini 2.5 Deep Think Model for AI Ultra Subscribers
Google officially launches the Gemini 2.5 Deep Think model for AI Ultra subscribers, boasting advanced reasoning capabilities.
Key Points
- • Gemini 2.5 Deep Think is available exclusively for AI Ultra subscribers at $250 per month.
- • The model uses parallel thinking techniques for enhanced problem-solving.
- • Deep Think achieved gold-medal standards at the International Mathematical Olympiad.
- • Performance benchmarks show it outperforms competitors like OpenAI’s O3 and Grok 4.
On August 1, 2025, Google unveiled its highly anticipated Gemini 2.5 Deep Think model, now available exclusively for subscribers of the Google AI Ultra plan, costing $250 per month. Designed to tackle complex reasoning tasks such as advanced mathematics and coding, Gemini 2.5 Deep Think significantly enhances performance over its predecessors, integrating advanced features developed from extensive tester feedback.
Deep Think's capabilities include a novel approach known as 'parallel thinking,' which allows the model to simultaneously evaluate and refine multiple ideas, improving problem-solving efficiency. This model has already achieved a gold-medal standard at the International Mathematical Olympiad (IMO) during initial testing, and public users can expect it to demonstrate bronze-medal capabilities, according to Google.
The introduction of Deep Think marks a significant advancement in AI technology, as it not only excels in traditional reasoning but also shows state-of-the-art performance across various benchmarks, outperforming competitors such as OpenAI's O3 and xAI’s Grok 4 in areas like coding and scientific knowledge. Users have noted improved content safety and tone-objectivity in Deep Think compared to its predecessor, the Gemini 2.5 Pro, although it tends to refuse more benign requests.
Access to Deep Think is facilitated through the Gemini app, where AI Ultra subscribers can utilize its features, including code execution tools and extended response options. However, use is limited to a fixed number of prompts each day; specifics on the limits remain undisclosed. Meanwhile, Google is exploring further usability through plans to release Deep Think via the Gemini API to a select group of trusted testers for development and enterprise applications.
In conclusion, the launch of Gemini 2.5 Deep Think underscores Google’s commitment to advancing AI technology and enhancing user experience, setting a new standard for reasoning capabilities in artificial intelligence models.