The AI takes on OpenAI's o1 reasoning variant. The model, dubbed s1, was trained using a dataset of 1,000 questions for under $50.
The Cohere CEO and former Google Brain member was impressed by DeepSeek's reasoning model, but doesn't think it'll be useful ...
DeepSeek AI and ChatGPT compared. DeepThink focuses on tasks requiring reasoning and deeper thinking, but its answers could be biased, and it stays mum sometimes!