OpenAI passes 1 billion weekly users as AI cost race tightens

The AI adoption number to watch is no longer only ChatGPT sign-ups. On July 31, 2026, OpenAI said its models now reach more than one billion active users and more than two million businesses. The company tied that scale to the previous day's GPT-5.6 price cuts: Luna is 80% cheaper, Terra is 20% cheaper, and Sol now has a Fast mode that can run up to 2.5 times faster than standard processing at twice the price.

The primary source is OpenAI's post, “Building abundant intelligence”. It frames the company's infrastructure strategy around a loop: broader adoption produces revenue and feedback, efficiency improvements lower the cost of service, and lower costs make more work economical. That framing matters because enterprise AI buyers increasingly compare models by the cost of completed tasks, not by token prices in isolation.

The post gives several concrete usage signals. OpenAI says that six months after signing up, people send roughly 50% more messages each day and use ChatGPT for about twice as many kinds of work. It also says ChatGPT Work is moving users from asking questions toward completing multistep work. Inside OpenAI, agentic work through Codex now accounts for 99.8% of weekly output tokens, with Finance named as one team where agentic tools have become a primary part of the workflow.

The cost side is just as important. OpenAI lists GPT-5.6 Luna at $0.20 per million input tokens and $1.20 per million output tokens after the 80% cut. GPT-5.6 Terra is listed at $2 and $12, respectively. Fast mode for GPT-5.6 Sol offers up to 2.5 times standard speed at twice the price, with no intelligence change. That pricing structure gives customers more room to route routine, latency-sensitive, and high-reasoning tasks differently instead of forcing one model into every workflow.

OpenAI also points to technical improvements behind the economics. It says GPT-5.6 Sol helped optimize production software used to serve its models, reducing end-to-end serving costs by 20%. It also helped improve speculative decoding, increasing token-generation efficiency by more than 15%. In a separate benchmark analysis cited by OpenAI, retained reasoning and context management raised GPT-5.6 Sol's public ARC-AGI-3 score from 13.3% to 38.3% while using six times fewer output tokens.

The open question is whether scale can keep outrunning infrastructure cost. One billion active users and two million businesses give OpenAI a distribution advantage, but lower prices also raise expectations that useful intelligence should become cheaper quickly. The next evidence to watch is not only user growth, but API consumption, enterprise retention, real task completion rates, and whether the cheapest GPT-5.6 tier can hold quality as workloads become more agentic.