Andre Saraiva’s Post

Today, we introduced our new o3 reasoning models, accompanied by some impressive benchmark performance results! 📈 The o3 model demonstrates significant gains in Math ∑, Coding 👨💻, and has beaten the previously established ARC-AGI benchmark 🤖 (which experts thought was still years away). It’s a leap forward that even leaves o1 far behind in performance. These advancements in reasoning capabilities are a testament to our commitment to pushing the boundaries of what’s possible in AI. For those who know me, it’s no secret that I’ve been deeply interested in planning and reasoning for a long time. I’ve been fortunate to contribute meaningfully to these advancements, and I’m incredibly excited to see how this progress can make a real difference in planning and reasoning applications. I’m incredibly proud of the progress we’ve made with o-series. If you haven’t yet explored the o1 model (not o1-preview) in your area of expertise, now is the time. Whether your focus is on developing new products, strategizing, designing physics experiments, interpreting complex research papers, or learning something entirely new, o1’s capabilities might just surprise you. For those dedicated to safety, we are offering early access to the o3 models. We’d love for you to apply and share your experiences through private benchmarks. You can find more details and the application link here: https://lnkd.in/dbu_yQmZ