OpenAI launches o3-pro, slashes o3 price by 80% in bid to widen AI lead

OpenAI has unveiled its most advanced AI model to date, the o3-pro, which surpasses competitors on key benchmarks and replaces the o1-pro. The o3-pro is now available for ChatGPT Pro and Team users, as well as through the developer API, with access for enterprise and education sectors beginning next week.

The model excels in math, science, and coding, setting a new standard for AI capabilities, as per the company’s release notes.

Alongside the rollout, OpenAI announced a steep 80% price cut for its o3 model, slashing rates from $10/$40 to $2/$8 per million input/output tokens. CEO Sam Altman confirmed the update on X, posting: Users will be “happy with o3-pro pricing for the performance.”

This aggressive pricing directly challenges the previous market dynamic, where Google’s Gemini 2.5 Pro, priced at $1.25–$2.50 for inputs and $10–$15 for outputs, had an edge on cost at comparable performance levels.

The strategic implications, however, go beyond the economics. “This is a market-defining move,” said Amandeep Singh, associate director at QKS Group. “It’s not pure commoditization, it’s ecosystem lock-in. OpenAI is lowering entry barriers, so enterprises standardize on its stack, much like AWS did in its early days.”

Manish Ranjan, research director at IDC, called the move a “game-changer” for startups and SMBs, making it feasible for them to scale their genAI prototypes into business-wide deployments. “This dramatically reduces the cost of entry while maintaining performance and agility,” Ranjan said, though he noted that many enterprises will still reserve o1-pro for high-stakes workloads where latency and reliability remain critical.

Enterprise appeal, and implementation friction

Early tests show o3-pro outperforms Google’s Gemini 2.5 Pro on AIME 2024 math benchmarks and Anthropic’s Claude 4 Opus on GPQA Diamond science tests.

Reviewers rated it best-in-class for science, education, programming, business, and writing. Its advanced capabilities, including web search, file analysis, and visual input reasoning, deliver reliable results in complex tasks like physics and coding, as per the company blog post.

Despite OpenAI’s aggressive pricing, mid-sized enterprises may struggle to capitalize. “The price drop opens the door, but it doesn’t build the hallway,” said Singh. “Cheap model access doesn’t guarantee scalable deployment. The real bottlenecks are orchestration, compliance, and governance.”

Singh also warned that without centralized AI policies, CSOs risk governance blind spots, such as prompt injection or data exposure. He noted that as o3-pro shares o3’s safety framework, SOC 2-compliant workflows are essential for regulated sectors.

He further cautioned against “prototype sprawl,” a proliferation of experimental GenAI projects that never reach production due to gaps in data readiness, AI talent, or evaluation tooling. Without a managed abstraction layer like “GenAI-as-a-Platform”, many organizations may be left behind, despite lower costs.

Ranjan said o3-pro offers enterprises more value for less, but only if they adapt to limitations like the lack of temporary chats, image generation, or Canvas support.

Orchestration, auditability, and model-agnostic platforms

With o3 serving three million business users and a 50% enterprise surge since February 2025, OpenAI’s pricing and performance are undeniably compelling. However, procurement priorities are shifting toward orchestration, auditability, and model-agnostic platforms. “You’re not just buying a model, you’re investing in adaptability and compliance,” Singh emphasized. Total cost of ownership now hinges on the infrastructure surrounding the model, not just token pricing.

While Ranjan thinks OpenAI will capture more market share, he advised CIOs to mix models, including open-source options for greater flexibility. Singh agreed, warning that without strong governance, organizations risk exposure and lock-in.

OpenAI, meanwhile, has teased more innovation, with a delayed open-weights model now expected later this summer. “Our research team did something extraordinary. It’ll be worth the wait,” Altman hinted on X. For now, OpenAI is betting on performance, pricing, and developer tools to maintain its edge. “o3-pro and the price cut mark an inflection point,” Singh concluded. “Competitive advantage will increasingly depend on governance, orchestration, and domain expertise, not just model quality.” As enterprises weigh cost, speed, and integration, OpenAI’s latest move signals a new phase in the race to define AI’s future.