OpenAI has announced a major milestone: the audience for its artificial intelligence services has surpassed 1 billion active users. Additionally, more than 2 million enterprises worldwide are now using the company's tools. Following this news, the developer of ChatGPT reported a significant reduction in the cost of access to its latest models.
Price Cuts for GPT-5.6
OpenAI has revised its pricing policy for two models in the latest GPT-5.6 family. Access to the GPT-5.6 Luna model has become 80% cheaper. The cost of processing is now $0.20 per 1 million input tokens and $1.20 per 1 million output tokens. Previously, these figures were $1 and $6, respectively.
The more powerful GPT-5.6 Terra model has also become 20% cheaper. Prices for it have settled at $2 for input and $12 for output tokens (previously $2.50 and $15). The cost of working with the flagship Sol model remains unchanged.
The price reduction was made possible thanks to internal optimization. Infrastructure maintenance costs decreased by 20%, while token generation efficiency increased by more than 15%. Thanks to improvements in context management, the GPT-5.6 Sol model showed significant performance growth in the ARC-AGI-3 benchmark: from 13.3% to 38.3%, while output token consumption decreased sixfold.
Competition and Economic Challenges
OpenAI CEO Sam Altman admitted that the price adjustment was a forced measure. Corporate clients have become more cautious regarding high AI implementation costs, demanding a clear understanding of economic returns. The situation was also influenced by active competition: Anthropic lowered prices for its Claude models, and cheaper Chinese alternatives have strengthened their position in the market.
For comparison: the mid-range model from competitors, Anthropic Claude Sonnet 4.6, costs $3 for input and $15 for output tokens, making it more expensive than OpenAI's Terra model.
Usage Dynamics and Financial Indicators
Analytics show an increase in user engagement. Six months after registration, clients start sending an average of 50% more messages per day and use ChatGPT to solve twice as many types of tasks. Notably, up to 99.8% of weekly output tokens are accounted for by Codex programming AI agents.
Despite the massive growth in the user base, OpenAI continues to operate at a loss. As of the end of 2025, the company reported revenue of $13.07 billion with a net loss of $38.5 billion.
