OpenAI Cuts API Prices 50%, Debuts GPT-6 Sol and Luna

Three distinct OpenAI moves landed in the past 48 hours: a major model launch with steep price cuts, a new math advisory structure, and a financial stake in a Silicon Valley education startup. Each tells a different part of the same story about where OpenAI is placing its bets.

Key points

• OpenAI launched two new GPT-6 models, Sol and Luna, cutting API prices by more than 50% to steer routine workloads to cheaper tiers.
• The company’s higher-complexity model, Astra, remains in place for demanding tasks, preserving a tiered product architecture.
• OpenAI’s AI has resolved more than 100 open math problems; a new advisory group now provides expert oversight of that accelerating research.
• OpenAI said it cannot slow or redirect its math research program, even as it adds external oversight.
• OpenAI joined Andreessen Horowitz, Anthropic, and Nvidia in backing a $42 million San Francisco program designed as a hands-on college alternative for the AI job market.

What the GPT-6 pricing move signals

The Sol and Luna launch is the more commercially significant announcement. Cutting API prices by more than half is not a gesture toward developers; it is a competitive repositioning. OpenAI is clearly responding to pressure from lower-cost inference providers, and the tiered structure, with Sol and Luna absorbing summarization and coding tasks while Astra handles complex workloads, is designed to capture volume at the low end without cannibalizing premium revenue.

For enterprise buyers, a 50%-plus price drop on routine inference materially changes unit economics. Companies that have been rate-limiting API calls for cost reasons now have less reason to hold back. That could accelerate adoption curves and deepen integration, both of which matter if OpenAI eventually pursues a public market transaction where revenue growth and customer retention are scrutinized closely.

The strategic risk is margin compression. OpenAI has not disclosed its infrastructure cost structure publicly, so it is unclear how much headroom exists below current pricing. Investors tracking the path to profitability should watch whether this move expands gross margins through volume or erodes them.

Does the math advisory group resolve the safety tension?

The math research development is harder to read. OpenAI’s AI resolving more than 100 open problems is a striking research milestone, and the formation of an expert advisory group looks like a reasonable governance response to rapid, potentially consequential progress.

The complication is OpenAI’s own stated position: it cannot slow or redirect the research. That framing raises a straightforward question about what the advisory group actually controls. Oversight bodies without the ability to pause or redirect carry limited authority in practice. This echoes a broader tension in OpenAI’s public posture, where safety governance structures are added alongside, rather than as a check on, accelerating capability development.

For outside observers, the honest read is that the advisory group provides expert input and external credibility, not a veto. Whether that is adequate oversight of AI resolving open mathematical problems at speed is a genuinely open question.

The education investment: branding or conviction?

The Horowitz Andreessen Academy announcement is smaller in dollar terms ($42 million total, with OpenAI as one of several backers), but it is worth understanding in context. OpenAI co-investing alongside Anthropic and Nvidia in a college alternative signals something about how these companies view the near-term talent pipeline. The program focuses on hands-on AI building and interpersonal skills, which suggests the backers believe the conventional four-year degree is not producing the kind of graduates they need fast enough.

For OpenAI specifically, participation keeps the company visible in workforce and education conversations at a moment when policymakers are paying close attention to AI’s labor market effects. It also positions OpenAI as a constructive actor in that debate, a useful posture for a company navigating regulatory scrutiny while pursuing a structural conversion to for-profit status.

The broader picture

Taken together, these three moves reflect a company pushing on multiple fronts simultaneously: competing aggressively on API pricing, accelerating frontier research while adding oversight with acknowledged limits, and investing in the talent ecosystem that will consume its products. None of these are contradictory, but the pace and breadth are notable. OpenAI is not consolidating; it is expanding its surface area across commercial, research, and educational domains all at once.