Google expands access to Gemini 2.5 Pro AI model, reveals competitive pricing for developers

Google has officially opened up wider access to its Gemini 2.5 Pro artificial intelligence model, moving the advanced AI system into public preview with increased rate limits via the Gemini API and Google AI Studio. The shift marks a strategic expansion in Google’s AI push, aimed at attracting developers with more flexible usage and competitive pricing.

The public preview follows a successful early-access rollout last month, where Gemini 2.5 Pro debuted as an experimental release with tight usage caps. Now, with demand surging, Google is increasing those limits and introducing a structured two-tier pricing model for developers.

Gemini 2.5 Pro: Two-Tier Pricing Explained

Developers using up to 200,000 tokens per month will pay:

    •    $1.25 (approx. ₹107) per million input tokens
    •    $10 (approx. ₹858) per million output tokens

For usage beyond the 200,000-token threshold, prices increase to:

    •    $2.50 (approx. ₹214) per million input tokens
    •    $15 (approx. ₹1,290) per million output tokens

These tokens can include text, images, audio, and video inputs, and the output cost accounts for high-performance reasoning capabilities. While the experimental version remains free with limited access, the public preview allows for more extensive deployment.

Google highlighted that these prices are well below competitors. For example:

    •    Anthropic’s Claude 3.7 Sonnet charges $3 per million input tokens and $15 per million output tokens.
    •    OpenAI’s GPT o1 model starts at $15 per million input tokens and climbs to $60 per million output tokens for outputs, though discounted rates are available for cached tokens.

The release of Gemini 2.5 Pro into public preview is part of Google’s ongoing effort to gain ground in the rapidly expanding generative AI market. The model is not yet available on Vertex AI, but Google confirmed its addition is imminent.

Touted by the company as its “most intelligent AI model to date,” Gemini 2.5 Pro supports multimodal inputs and is optimised for complex reasoning features designed to rival and potentially undercut OpenAI and Anthropic’s top-tier offerings.