OpenAI Surpasses One Billion Users, Boosted by the Drop in GPT-5.6 Luna Prices (-80%)

On July 30, 2026, in a strategic post that went somewhat unnoticed amid product announcements, OpenAI revealed two figures and an action that, combined, reshape the landscape of consumer AI:

• OpenAI models now exceed one billion active users
• More than 2 million business clients integrate them
• The API price for GPT-5.6 Luna drops by 80% and for GPT-5.6 Terra by 20%

The message is clear: OpenAI is willing to sacrifice a considerable margin to retain its user base against the Chinese wave (DeepSeek V4, Kimi K3, Qwen) and the aggressive offerings from Anthropic and Google.

The New API Prices

The GPT-5.6 pricing structure includes three tiers — Luna, Terra, Sol — in increasing order of capacity. The new rates are:

• Luna: $0.20 / million input tokens, $1.20 / million output tokens (compared to $1 and $6 before — –80%)
• Terra: $2 / million input tokens, $12 / million output tokens (compared to $2.50 and $15 — –20%)
• Sol: unchanged price. The flagship remains premium.

This is the second time in 12 months that OpenAI has lowered prices for its mid-range lineup. The strategic direction is clear: expand economic usage, maintain consumer revenue through Sol and ChatGPT subscriptions.

Why Now?

Three pressures converge:

1. China is slashing prices on video and LLMs

ByteDance Seedance 2.5 is up to 7× cheaper than Sora for video. DeepSeek V4 Flash and Kimi K3 offer unbeatable value for money on LLMs. In the face of this pressure, OpenAI can no longer maintain 2024 API prices.

2. The B2B market is becoming cost-sensitive

Companies that have integrated ChatGPT and the API into their workflows are starting to receive five- or six-figure monthly bills. The IT department is making decisions. A model at 80% of the price for 90% of the performance becomes much more attractive than a "better" model that's 5× more expensive.

3. GPT-5.6 is natively cheaper to run

OpenAI explicitly attributes the reductions to internal efficiency gains achieved during the development of GPT-5.6: production software optimization, improvement in speculative decoding. The company is passing some of these savings back to the market — and keeping the rest.

The Billion Users: What It Represents

To put it in context, one billion active users is:

• More than the global user base of LinkedIn (~1 billion)
• Half of TikTok (~2 billion)
• About 1 in 8 humans on Earth

No AI company has ever reached this scale. In comparison, Anthropic claims a few tens of millions of users on Claude, Google Gemini a few hundred million boosted by Android integration — but no player has crossed the symbolic billion mark.

The figure of 2 million businesses also deserves attention. It's an order of magnitude above the client base of Microsoft Copilot for Business or the enterprise base of Google Workspace integrated with Gemini. OpenAI captures an entire segment of the B2B market.

The Strategic Reading

OpenAI is accepting a short-term margin squeeze for two objectives:

1. Lock in the ecosystem. Every company that integrates GPT-5.6 Luna into its workflow becomes harder to dislodge in 12 months — even if a Chinese competitor arrives at 50% of the price.
2. Fund ASTRA. OpenAI's next major model (likely GPT-6, see the announcement of ten mathematical breakthroughs) will require colossal training budgets. A massive, paying client base — even with reduced margins — remains the primary cash flow driver.

The Risks

Three areas of concern deserve attention:

• Profitability. OpenAI is still not profitable. Lowering prices by 80% on a major consumer tier without massively scaling compresses results.
• Race to the bottom. Anthropic (Claude Sonnet), Google (Gemini 3.5 Flash), and Chinese players will align or attempt to undercut. The unit margin of the entire industry is contracting.
• Customer support. A billion users multiply edge cases, incidents, GDPR requests, disputes. OpenAI had to massively recruit in compliance and support in 2025-2026, with non-trivial costs.

Key Takeaway

OpenAI is no longer a research lab with peripheral monetization. It has become the largest distributor of artificial intelligence in the world, with a user base on par with major social networks. Pricing decisions that were once technical adjustments are now macro-economic moves that drive an entire sector.

Having crossed the billion mark, the question is no longer "Will OpenAI succeed in becoming mainstream?" It is: how long can it maintain this position against competitors building comparable models at a fifth of the production cost? The answer will come in the coming quarters.