# OpenAI Hits 1 Billion Users as It Slashes GPT-5.6 Luna Prices 80%

Page: https://hirantha.me/blog/openai-hits-1-billion-users-as-it-slashes-gpt-56-luna-prices-80
Published: 2026-08-05
Written by: hirantha.me Newsroom (AI-generated). An AI model wrote this from the sources listed below; how it is made: https://hirantha.me/about/newsroom
Topics: openai, chatgpt, gpt-5.6, ai-pricing, deepseek

> OpenAI says its tools now reach over 1 billion users, a day after cutting GPT-5.6 Luna prices 80% to compete with cheap open-weight rivals.

## A round number with real numbers behind it

OpenAI said on July 31, 2026 that its products now reach more than 1 billion active users and more than 2 million businesses. The company's own framing leaned on the comparison: Facebook took six years to reach 1 billion users, while OpenAI got there in under four years from ChatGPT's November 2022 launch. The figure isn't limited to the ChatGPT chat window. It covers OpenAI's full lineup of consumer and business interfaces, including Codex, the company's AI coding tool, and ChatGPT Work, which lets people reach outside applications through OpenAI's products.

"Our models now reach more than one billion active users and more than two million businesses. As people gain confidence in the technology, they use it more deeply," the company said in its announcement.

The milestone arrived later than OpenAI originally expected. The company had hoped to cross 1 billion earlier in 2026, after reporting 900 million weekly active users in late February. Growing competition from rival model providers, including a wave of inexpensive Chinese open-weight models, appears to have slowed that climb.

## The more interesting news is what happened the day before

A day before the user milestone, on July 30, OpenAI cut prices on two of the three tiers in its GPT-5.6 model family. GPT-5.6 Luna, the smallest and fastest model in the lineup, dropped by 80% across every pricing category: input tokens fell from $1.00 to $0.20 per million, cached input tokens fell from $0.10 to $0.02 per million, and output tokens fell from $6.00 to $1.20 per million. GPT-5.6 Terra, the mid-tier model, got a smaller 20% cut, with input dropping from $2.50 to $2.00 per million tokens, cached input from $0.25 to $0.20, and output from $15.00 to $12.00.

GPT-5.6 Sol, the flagship model, kept its existing price but gained a new option instead: a Fast mode that delivers up to 2.5 times the throughput of standard processing, at twice the standard price, or $10 per million input tokens and $60 per million output tokens. OpenAI describes this as priority compute access rather than a smarter model, meaning Sol's underlying intelligence is unchanged.

OpenAI attributed the price cuts to efficiency work rather than a defensive move on its own, saying it used GPT-5.6 to optimize its own production infrastructure. The company cited a 20% reduction in serving costs from improved graphics processing unit (GPU) kernels and more than 15% better token-generation efficiency from improved speculative decoding, a technique where a smaller draft model predicts likely next tokens for a larger model to verify. As a sign of how much OpenAI trusts those savings, it also moved its own internal auto-review tooling in ChatGPT and the Codex command-line interface from the older GPT-5.4 model to Luna, which it says cuts that internal cost by roughly 90%.

## Why now

The framing of self-optimization aside, the timing tracks closely with pricing pressure from competitors. Cheap open-weight models, DeepSeek chief among them, had pushed Luna's original pricing out of step with what buyers can get elsewhere. Chinese-origin models have reportedly taken as much as 46% of US enterprise token usage on the OpenRouter marketplace at points this year, occasionally overtaking US-built models in that measure.

Even after the cut, Luna isn't the cheapest option on the market. Combining its new input and output rates puts it below Google's Gemini 3.5 Flash-Lite and well below the pricier Gemini 3.6 Flash, but Chinese flash-tier models such as DeepSeek's own low-cost offering and Xiaomi's MiMo-V2.5 Flash still undercut it on raw per-token cost. What the cut does is move Luna from clearly overpriced relative to its closest rivals to competitive with most of them, while keeping OpenAI's brand and tooling ecosystem attached.

## What it means in practice

For anyone building on OpenAI's application programming interface (API), the practical change is straightforward: Luna is now a viable default for high-volume, latency-sensitive workloads where GPT-5.6 Sol would have been overkill, and it's worth re-checking whether tasks currently routed to Terra or Sol could run acceptably on Luna at a fraction of the cost. For workloads where speed matters more than price, Sol's new Fast mode is a lever worth testing before reaching for a larger model.

The billion-user figure matters less for what it changes today than for what it signals about the market OpenAI is now competing in: one large enough that even single-digit percentage shifts in where developers route their API traffic are worth an 80% price cut to defend. Expect further price moves from OpenAI and its rivals as open-weight Chinese models continue setting the floor on cost per token.

## Questions

**What does OpenAI's 1 billion user figure actually count?**
It covers active users across all of OpenAI's interfaces, including ChatGPT, the Codex coding tool, and ChatGPT Work, not just the main chat app.

**How much cheaper is GPT-5.6 Luna now?**
Luna's price dropped 80% across the board: input tokens fell from $1.00 to $0.20 per million and output tokens fell from $6.00 to $1.20 per million.

**Did GPT-5.6 Sol get cheaper too?**
No, Sol's price is unchanged. Instead it gained an optional Fast mode that runs at up to 2.5 times the speed for twice the price, with no change to the model's intelligence.

**Why did OpenAI cut prices now?**
OpenAI attributes it to internal efficiency gains, but the cuts follow pressure from cheap open-weight Chinese models like DeepSeek's, which had made Luna's old pricing uncompetitive.

**Is Luna the cheapest model available now?**
No. It now undercuts Google's Gemini 3.5 Flash-Lite and Gemini 3.6 Flash, but some Chinese flash-tier models, including DeepSeek's and Xiaomi's MiMo-V2.5 Flash, still cost less per token.

## Sources

- [OpenAI says has more than 1 billion active users](https://www.bnnbloomberg.ca/business/artificial-intelligence/2026/07/31/openai-says-has-more-than-1-billion-active-users/), BNN Bloomberg
- [Announcing a major Price drop for 5.6 Terra and Luna and Fast mode for 5.6-Sol](https://community.openai.com/t/announcing-a-major-price-drop-for-5-6-terra-and-luna-and-fast-mode-for-5-6-sol/1388484), OpenAI Developer Community
- [OpenAI Slashes GPT-5.6 Luna Prices 80%, Undercutting DeepSeek as AI Price War Intensifies](https://mlq.ai/news/openai-slashes-gpt-56-luna-prices-80-undercutting-deepseek-as-ai-price-war-intensifies/), MLQ News