On Tuesday, June 10, 2025, OpenAI announced the launch of its latest and most capable model to date, o3-pro.
Paired with this release, the company revealed a staggering 80% price reduction for its flagship reasoning model, o3, opening up new avenues for innovation in the advanced AI field.
o3-pro: Superior Reasoning for Complex Challenges
o3-pro stands as an enhanced and upgraded version of the o3 model that OpenAI introduced earlier this year.
This new model was specifically engineered to tackle complex problems step-by-step, an approach that boosts its reliability and performance in precise fields like physics, mathematics, and programming.
According to OpenAI, human evaluators showed a clear preference for o3-pro over its predecessor, o3, across all testing categories. The model particularly excelled in critical domains such as science, education, programming, business, and writing assistance. Evaluators also noted its consistent superiority in output clarity, comprehensiveness, and its ability to follow instructions accurately.
Features and Performance
The o3-pro model can access a suite of tools that expand its capabilities. Its toolkit includes web search, file analysis, visual input comprehension, Python execution, and response personalization based on accumulated memory.
OpenAI o3-pro has access to tools that make ChatGPT useful—it can search the web, analyze files, reason about visual inputs, use Python, personalize responses using memory, and more.
— OpenAI (@OpenAI) June 10, 2025
Internal benchmarks conducted by the company show o3-pro achieving impressive results on well-known AI performance metrics, outperforming prominent competing models in areas of advanced mathematics and specialized scientific knowledge.
Pricing
API pricing for o3-pro starts at $20 per million input tokens and $80 per million output tokens.
Despite its superior capabilities, OpenAI noted that the model's response times may be longer compared to other models like o1-pro.
Current limitations also include a lack of support for temporary chats, no image generation capabilities, and incompatibility with the Canvas workspace.
Some developers with early access have observed that o3-pro is significantly smarter but requires more context to deliver optimal results. When provided with that context, it can produce highly accurate and impactful plans and analyses.
o3 Price Reduction: Making Advanced AI More Accessible
OpenAI's announcements were not limited to the o3-pro launch; they also included a radical 80% price reduction for the o3 model.
CEO Sam Altman stated that this price cut aims to encourage developers to experiment more broadly with the model.
Under the new pricing structure, using o3 now costs $2 per million input tokens and $8 per million output tokens, with an additional discount for cached information.
This move positions the o3 model to be highly competitive against other advanced reasoning models, such as Google DeepMind's Gemini 2.5 Pro and Anthropic's Claude Opus 4, as well as some lower-cost models like those from DeepSeek.
Furthermore, OpenAI is now offering a flexible batch processing mode for o3 at $5 for input and $20 for output, giving developers greater control over cost and latency depending on their specific tasks.
Such a significant cost reduction makes the advanced reasoning capabilities of o3 accessible to a wider range of developers, startups, and research teams who may have previously found the cost of sophisticated models to be a barrier.
Ultimately, these announcements from OpenAI reflect a dual strategy:
First, to push the boundaries of innovation by introducing more powerful and intelligent models like o3-pro.
Second, to make these advanced technologies more accessible and affordable, which will undoubtedly accelerate the pace of development and adoption of AI solutions across various sectors.
