On Friday, Google released the API pricing for the Gemini 2.5 Pro, an AI inference model with industry-leading performance in several benchmarks measuring coding, inference and mathematics.
Prompts up to 200,000 tokens cost $1.25 per Gemini 2.5 Pro input token (approximately 750,000 words, the entire “Rod of the Rings” series) and $10 per million output tokens. For prompts over 200,000 tokens (which most Google competitors don’t support), the Gemini 2.5 Pro costs $2.50 per million input tokens and $15 per million.
This price will be more expensive for the Gemini 2.5 Pro for developers than other AI models currently offered by Google, including Gemini 2.0 Flash ($0.10/MILLEN INPUT TOKENS, $0.40/million output tokens). The Gemini 2.5 Pro also makes it more expensive than several other frontier AI models, such as Openai’s O3-Mini ($1.10/Million Input Token, $4.40/Million Output Token) and Deepseek’s R1 ($0.55/Million Input Token, 2.19 million/Million Output Token).
To be fair, the Gemini 2.5 Pro, available for free with strict rate limits, comes at a cheaper price than other highly competitive models, such as Anthropic’s Claude 3.7 Sonnet ($3 million/million input token, $15 million/million output token), and Openai’s GPT-4.5 (75/million input token, $150/1.5 million/million output token). The tech industry’s initial response is largely positive, and it praises what developers think is a wise rate.
But roughly speaking, there appears to be some upward pressure on pricing for flagship models. The costs of recent top releases from labs like Google, Openai, Anthropic have risen, not down. For example, take a look at the O1-Pro, which opened recently. This is the company’s most expensive API product, with $150/million input tokens and $6 million/million output tokens.
High demand could drive the trends. According to Google CEO Sundar Pichai, the Gemini 2.5 Pro is the most in demand AI model of developers, with an 80% increase in usage on Google’s AI Studio platform and Gemini API.
Source link
