Tag
GPT-6.2 Astra is being showcased as a strong robotics foundation model, having risen to the #1 spot in RoboColosseum's SO-101 Elo rankings after hundreds of head-to-head comparisons in a single day.
Someone compared GPT-6.2 Astra and Sol across tasks including 3D scenes, games, slideshows, data visualization, websites, and interactive apps, highlighting Astra's advantages in detail and precision alongside Sol's strengths in motion design, with the cost of each run during the recording included.
GPT-6.2 Astra independently proved a 20-year-old mathematical theorem in 94 minutes, bypassing three key results of the original proof, with its 1,794-line proof formally verified without human intervention.
A user pushes back on Twitter influencers calling GPT-6.2 Astra the worst model ever, arguing it outperforms Fable 5.2 at chess and solves novel mathematical proofs, urging people to test models themselves.
OpenAI announced the release of GPT-6.2 Astra and Sol, described as efficient real-world performers, while users debate whether the update meets expectations compared to prior models like Fable 5.2.
Sam Altman announces the release of GPT-6.2 Sol and Astra models, highlighting their capabilities and token efficiency, with 6.2 Sol ultrafast now expanded to Pro 200 plans.
OpenAI 发布 GPT-6.2 Astra(旗舰前沿模型)与 GPT-6.2 Sol(高效日常模型),两者在更少输出 token 下取得更好结果,并在 Agents' Last Exam、Terminal-Bench-Science 等基准上刷新纪录,同时 API 定价更低。
OpenAI showcases its GPT-6.2 Astra and GPT-6.2 Sol models in a video styled as a short office skit, highlighting their capabilities in real-world work scenarios such as accounts payable processing, financial calculations (net debt and earnings per share), and presentation creation.