OpenAI rolls out o1 reasoning model in its API for certain developers

There is no denial that 2024 has been the year of artificial intelligence. We"ve witnessed the emergence of tools like Sora, SearchGPT, and Gemini 2 from major tech players. Recently, OpenAI joined the fray with its latest model, o1. According to OpenAI, o1 excels at tackling complex tasks such as mathematics, coding, and scientific research, focusing on the heavy-duty intellectual stuff. However, don"t expect o1 to browse the web or analyze images; it"s laser-focused on high-level reasoning.

Now, OpenAI has announced that this "super-smart" reasoning model is now available through its API. But there"s a small catch: it"s currently accessible only to certain developers. Specifically, o1 is rolling out to developers in OpenAI"s "tier 5" usage category. To qualify for tier 5, developers must have spent at least $1,000 on the OpenAI platform with an account that"s more than 30 days old since their first payment.

This release replaces the previous o1-preview model in the OpenAI API. OpenAI states that o1 is production-ready, offering features such as:

  • Function calling: Seamlessly connect o1 to external data and APIs.
  • Structured Outputs: Generate responses that reliably adhere to your custom JSON Schema.
  • Developer messages: Specify instructions or context for the model to follow, such as defining tone, style, and other behavioral guidance.
  • Vision capabilities: Reason over images to unlock many more applications in science, manufacturing, or coding, where visual inputs matter.
  • Lower latency: o1 uses on average 60% fewer reasoning tokens than o1-preview for a given request.
  • A new reasoning_effort API parameter: Allows you to control how long the model thinks before answering.

These enhancements aim to provide developers with a more efficient and versatile tool for complex problem-solving. OpenAI"s o1 represents a significant advancement in AI reasoning capabilities, aligning with the industry"s shift towards "long thinking" AI models that deliberate more deeply before responding.

While the initial access is limited, OpenAI plans to expand availability in the future, potentially allowing a broader range of developers to leverage o1"s advanced reasoning skills.

Report a problem with article
Next Article

Gemini Advanced subscribers can now try out Gemini-Exp-1206 on the web

Previous Article

Indiana Jones and the Great Circle update targets Xbox lighting issues, DLSS bugs, more