OpenAI slashes GPT‑5.6 prices by up to 80% as enterprise budgets tighten

OpenAI announced a sweeping price reduction for its GPT‑5.6 family on July 30, trimming the Luna model’s cost per token by 80% and the Terra model’s input and output rates by 20%. The company also doubled the speed of its premium Sol model, making it 2.5 times faster in the API.

The cuts arrive as corporate AI budgets surge, with firms like Uber and Microsoft reporting that their allocations are quickly exhausted. Many enterprises lack visibility into the true operational cost of running large language models, often underestimating expenses that can exceed human labor costs.

Luna, the fastest and least expensive GPT‑5.6 variant, is now positioned for high‑volume routine work that does not demand deep knowledge. Terra, designed for everyday knowledge tasks, offers a modest price drop that could broaden its appeal to smaller teams. Sol, the flagship model, remains unchanged in price but gains a significant performance boost, potentially offsetting higher usage costs for power users.

The announcement follows Microsoft’s recent earnings call, where the company signaled a shift from a cost‑per‑user model to a cost‑per‑user‑plus‑consumption framework. That change would raise enterprise AI spending for teams deploying multiple agents per employee.

OpenAI’s pricing strategy reflects a broader industry trend toward making advanced AI more affordable and accessible. The company’s stated goal is to “make advanced intelligence more abundant, affordable, and useful.” By lowering entry barriers, OpenAI hopes to expand its user base while maintaining profitability through higher volume.

The price cuts also underscore the growing scrutiny of AI costs. As models grow larger—Kimi K3, for example, boasts 2.8 trillion parameters—computational demands strain infrastructure. Chinese startup Moonshot AI had to pause new subscriptions after demand outpaced capacity, illustrating the challenges of scaling high‑performance models.

From a historical perspective, AI pricing has fluctuated as hardware costs and model complexity rise. Early generative models were priced by token count, but as inference becomes more efficient, providers can offer discounts without sacrificing revenue. The current environment, however, forces a reevaluation of cost structures, especially for enterprises that rely on multiple AI agents.

For practitioners, the key takeaway is that model choice now hinges on both cost and speed. Luna’s low price may be attractive for bulk data processing, while Sol’s speed upgrade could justify its premium for latency‑sensitive applications. Enterprises must also invest in cost‑tracking tools; without clear visibility, budgets can balloon unexpectedly.

Looking ahead, the question is whether other vendors will follow OpenAI’s lead. If price wars erupt, we may see a shift toward subscription‑based or usage‑tiered models that offer predictable budgeting. The industry’s response will shape how quickly advanced AI becomes a mainstream business tool.

FAQ
1. What is GPT‑5.6? GPT‑5.6 is OpenAI’s latest large‑language model, offering improved reasoning and contextual understanding over its predecessors.
2. How much did the prices drop? Luna’s cost fell by 80%, Terra’s by 20%, and Sol’s speed increased 2.5× without a price change.
3. Why are AI costs rising? Larger models require more compute, and enterprise usage patterns—multiple agents per user—drive higher consumption.
4. How can enterprises track AI spend? Implementing tooling that aggregates token usage across APIs and correlates it with billing data is essential for budget control.

Forbes
ABC News
Evertune
Forbes