o3-mini and o3-mini-high details leaked before OpenAI’s announcement

OpenAI is expected to announce the release of its new reasoning AI models, o3-mini and o3-mini-high, as part of the o3 family. These models aim to enhance problem-solving capabilities in areas like coding, science, and mathematics while offering faster response times compared to their predecessors. The announcement aligns with prior hints from OpenAI CEO Sam Altman and updates discovered in the latest ChatGPT web build, which includes a pop-up introducing these models.

o3-mini announcement

The o3-mini family introduces two versions: o3-mini and o3-mini-high. Both are optimized for reasoning-intensive tasks but differ in compute settings for performance flexibility. The models reportedly outperform the o1 pro model in speed and, in some cases, match or exceed its reasoning capabilities. A key feature is their ability to handle complex queries with reduced latency, making them suitable for both casual and professional use. Users will encounter a pop-up in ChatGPT with an option to try these models directly, maintaining the same user interface as the current o1 experience.

About OpenAI

OpenAI has been iterating on its reasoning models since the release of o1 in September 2024. The o3 series represents a significant step forward, focusing on advanced reasoning benchmarks such as ARC-AGI and SWE-Bench Verified. While the full o3 model targets high-compute applications, o3-mini caters to users needing cost-effective yet powerful AI solutions. This diversification reflects OpenAI's strategy to balance accessibility with premium offerings, as Plus subscribers will enjoy higher usage limits.

Timeline and Accessibility

The official rollout of o3-mini is anticipated today, January 31, 2025, following Altman’s earlier confirmation of a late-January release. The model will be available across both free and paid ChatGPT tiers, with free-tier users facing stricter query limits. Additionally, API access will be launched simultaneously for developers.