OpenAI has introduced o3-mini, a new reasoning model designed to deliver faster responses while using less compute than larger systems. The company says the model improves on its predecessor in math and programming tasks, making it a more efficient option for developers and product teams.

According to OpenAI, o3-mini is built to balance speed, quality, and cost. The company claims the model reduces inference costs by 40%, a change that could make advanced AI features more practical for wider deployment across apps and services.

The release reflects the growing industry push toward smaller models that can still handle complex reasoning work. As AI tools spread across search, coding, and workplace software, performance gains paired with lower operating costs are becoming a key competitive advantage.

OpenAI has positioned o3-mini as part of that shift, targeting users who need stronger technical performance without the expense of larger frontier systems. The model’s benchmark gains in mathematics and coding suggest the company is continuing to refine efficiency as much as capability.