OpenAI Says Its Services Now Used by Over a Billion People and More Than Two Million Companies

OpenAI reported that its AI models are now used by over 1 billion active users and more than 2 million enterprises. The company recently reduced prices on two models of its latest GPT-5.6 family: access to GPT-5.6 Luna dropped by 80%, to $0.20 per 1 million input tokens and $1.20 per 1 million output tokens (previously $1 and $6); GPT-5.6 Terra dropped by 20%, to $2 and $12 for input and output tokens, respectively (previously $2.50 and $15). The flagship model Sol remains unchanged in price. The price reductions were possible due to efficiency gains in GPT-5.6 development: service costs fell by 20%, and token generation efficiency improved by more than 15%. Context management improvements helped GPT-5.6 Sol increase its ARC-AGI-3 benchmark score from 13.3% to 38.3% while using six times fewer output tokens. American labs are forced to adjust prices because corporate clients are wary of high AI costs; pricing became a key issue, said OpenAI CEO Sam Altman. The ChatGPT developer had discussed price cuts for several weeks, concerned that Anthropic lowered prices for its Claude models. Corporate clients became reluctant to allocate large budgets to AI without a clear economic return, and competition from cheaper Chinese models intensified. Meanwhile, Anthropic's mid-tier Claude Sonnet 4.6 costs $3 per 1 million input and $15 per 1 million output tokens, which is more expensive than OpenAI's Terra. Six months after registration, users send about 50% more messages per day and use ChatGPT for roughly twice as many task types. Up to 99.8% of weekly output tokens go to the Codex AI programming agent. OpenAI continues to operate at a loss: in 2025 it showed $13.07 billion in revenue with a net loss of $38.5 billion.