OpenAI Makes Its New o3-pro Model Available for ChatGPT Pro Users
OpenAI is making its o3-pro model available for Pro users in ChatGPT, as the company writes in a blog post. The o3-pro model is designed to use and combine all tools that ChatGPT has at its disposal. Image generation and the use of Canvas are not supported; for these, the company refers to its other models.
Access to tools
o3-pro, like o3, has access to various tools. It can search the internet, analyze files, reason about visual input, use Python, and personalize responses.
read also
OpenAI Introduces New AI Models: o3 and o4-mini
OpenAI emphasizes that due to the use of these tools, responses with these models may take slightly longer compared to o1-pro. The company also recommends using the model for challenging questions where reliability takes precedence over speed.
O3-Pro model
OpenAI compares its models o1-pro, o3(medium), and o3-pro based on several benchmarks. For instance, the new model scores 90 percent on the AIME 2024 benchmark, which is about ten percent higher compared to the two other models. Furthermore, the o3-pro model also performs better on the GPQA Diamond and Competition Code benchmarks.
Restrictions
“At the moment, temporary chats are disabled for o3-pro as we are resolving a technical issue”, according to OpenAI. The company does not share details about the specific technical problems. Moreover, image generation is not supported in o3-pro. For this, users can fall back on the GPT-4o, OpenAI o3, or OpenAI o4-mini models. Finally, Canvas is currently not supported within o3-pro.