OpenAI drops o3 and o4-mini and they can think in images
OpenAI has launched two new AI reasoning models, o3 and o4-mini, designed to pause and work through questions before responding. The company announced the launch on Wednesday, making the models available to subscribers of its Pro, Plus, and Team plans.
O3 is touted as OpenAI’s most advanced reasoning model, outperforming its predecessors in tests measuring math, coding, reasoning, science, and visual understanding capabilities. O4-mini, on the other hand, offers a competitive balance between price, speed, and performance. Both models can generate responses using tools in ChatGPT, such as web browsing, Python code execution, image processing, and image generation.
The new models are part of OpenAI’s effort to stay ahead in the global AI race, competing with companies like Google, Meta, and Anthropic. O3 achieved state-of-the-art performance on SWE-bench verified, scoring 69.1%, while o4-mini scored 68.1%. OpenAI’s previous best model, o3-mini, scored 49.3% on the same test.
OpenAI claims that o3 and o4-mini are its first models capable of “thinking with images.” Users can upload images to ChatGPT, and the models will analyze them during their “chain-of-thought” phase before responding. The models can also run Python code directly in the browser via ChatGPT’s Canvas feature and search the web for current events.
Image: OpenAI
In addition to ChatGPT, the models will be available via OpenAI’s developer-facing endpoints, the Chat Completions API and Responses API. Pricing for o3 is $10 per million input tokens and $40 per million output tokens, while o4-mini is priced at $1.10 per million input tokens and $4.40 per million output tokens.
OpenAI plans to release o3-pro, a version of o3 that uses more computing resources, exclusively for ChatGPT Pro subscribers in the coming weeks. CEO Sam Altman has hinted that o3 and o4-mini may be the company’s last standalone AI reasoning models before GPT-5, a unified model that combines traditional and reasoning models.