OpenAI Expands Access to Advanced o1 Reasoning Model for Developers

The full o1 model, designed for complex reasoning, is now available through OpenAI's API for select high-tier developers with new features and pricing.

OpenAI has made its most advanced o1 reasoning model available to Tier 5 developers via its API, requiring a $1,000 monthly spend and accounts older than 30 days.
The o1 model introduces significant improvements over its preview version, including enhanced accuracy, visual reasoning, function calling, and a customizable reasoning_effort parameter.
Developers can use o1 for tasks like coding, mathematics, and visual analysis, with benchmarks showing major performance gains in areas such as SWE-bench Verified and AIME tests.
The model's pricing reflects its high computational demands, with costs of $15 per 750,000 words analyzed and $60 per 750,000 words generated—substantially higher than GPT-4o models.
OpenAI also announced updates to its Realtime API, including WebRTC integration for low-latency voice applications, new fine-tuning methods, and SDKs for Go and Java developers.