GPT-5.6 Sol Launch: Emergency Policy Changes, 8M Users, and the AI Cost Revolution

OpenAI also temporarily reduced a context window setting from 372,000 to 272,000 tokens after discovering the original size contributed to infrastructure strain, and fixed plugin submission issues and default reasoning mode access .

The 8 Million Active User Milestone: What the Number Really Means

By July 14–15, 2026, the combined weekly active users across Codex and ChatGPT Work reached 8 million . This represents a growth rate that is nearly unprecedented in enterprise SaaS: more than 7x expansion in five months, up from approximately 1 million weekly active users in February 2026 .

Important caveat: The 8 million figure is a combined total for Codex and ChatGPT Work, not a standalone Codex number . Codex engineering lead Tibo Sottiaux confirmed this on X . Prior to the GPT-5.6 launch, Codex alone had reached 5 million weekly active users in June 2026 . The growth accelerated sharply after general availability: the user base went from 6 million to 8 million in just three days .

Efficiency Improvements: The Three-Tier Pricing Revolution

Perhaps the most strategically significant aspect of the GPT-5.6 release is the new three-tier model architecture, which offers dramatically different price-performance tradeoffs:

Pricing confirmed across OpenAI's official help center and multiple independent sources .

• Terra is the headline efficiency story: it delivers GPT-5.5-class performance at roughly half the cost of GPT-5.5 ($5/$30) . It is now the default for Free and Go users in Work and Codex .
• Luna is the under-discussed win: a "frontier-adjacent" model at $1/$6 per million tokens—the cheapest tier—while still clearing 82% on Terminal-Bench .
• Sol holds the same price as GPT-5.5 ($5/$30) but delivers state-of-the-art results across coding, knowledge work, cybersecurity, and science with fewer tokens and lower estimated cost than competing models .

OpenAI also introduced new reasoning modes: max, available to all GPT-5.6 users in Codex, and ultra, which deploys multi-agent acceleration for the hardest professional tasks on Plus and higher plans .

Broader Competitive Landscape

The GPT-5.6 launch did not happen in a vacuum. The competitive dynamics among the top three AI labs have shifted significantly:

• Anthropic overtook OpenAI in U.S. enterprise AI spending for the first time in 2026, driven by superior performance in coding agents and long-context enterprise workflows . Anthropic also released new enterprise controls for Claude in early July 2026 .
• Google (Gemini 2.5 Pro) competes with a deeply integrated Workspace ecosystem and tied OpenAI at 74 in overall competitive scoring across nine weighted categories .
• The three providers—OpenAI, Anthropic, and Google—together represent the overwhelming majority of enterprise LLM usage globally .
• OpenAI proposed a novel equity-sharing structure for government involvement, signaling a shift in how it engages with regulators .

Regulatory Context: Government Oversight and the EU AI Act

GPT-5.6 Sol's path to public availability was anything but standard:

• 13-day government-restricted preview: Starting June 26, 2026, access was limited to a select group of trusted partners and organizations before receiving U.S. government approval for general availability . OpenAI stated it does not want this kind of prior government access to become the norm but accepted it as a transitional step .
• EU AI Act deadline extension: The EU struck a compliance deal pushing the deadline for high-risk AI rules from August 2026 to December 2027, giving companies deploying AI in regulated areas (hiring, lending, education, law enforcement, medical devices) 16 additional months to comply .

What Happens Next

The first week of GPT-5.6 Sol demonstrated that demand for frontier AI capabilities is not slowing down—it is accelerating faster than even OpenAI anticipated. The emergency usage resets and tier changes were stopgap measures; larger improvements are expected, including better visibility on usage and reset timing, and clearer guidance on when to use ChatGPT Work versus Codex . Meanwhile, the pricing revolution introduced by the Sol/Terra/Luna tiers has put competitive pressure on rivals, forcing the entire market to rethink the cost of intelligence.