top of page

OpenAI o1

1440 Daily Digest

Artificial intelligence startup OpenAI released its latest flagship large language model yesterday, dubbed OpenAI o1, its first major product debut since the May reveal of GPT-4o. The new model and a more cost-effective, limited version are currently only in preview for paying users and have a weekly message limit. 


OpenAI o1—previously hyped as the Strawberry model—reportedly differs from other models by the strength of its "reasoning" capability. OpenAI says the model breaks down queries into discrete steps before offering an answer, privately refining the response (watch demo). The model was trained via reinforcement learning, an approach that teaches the system via rewards and penalties. The text-based model surpassed GPT-4o on math and coding abilities and cannot currently access the web.


Over 200 million people use ChatGPT products monthly, and the company boasts nearly 60% of the market share in AI search. The product debuts as OpenAI announced a new fundraising round valuing it at $150B Wednesday. Watch how generative AI works here

Comments


  • White Facebook Icon

© 2024 389 Country. Powered by Talt Multi Media

bottom of page