Yesterday, OpenAI released two new models, o1-preview and o1-mini, codenamed “Strawberry.” These models have been trained to spend time reasoning before responding: first planning the steps necessary to solve a problem, then iteratively executing them.
To date, a collaborative approach has been necessary to get the best results from AI. Chain of thought encourages multistep thinking, but everything is done in one pass—the LLM doesn’t have the opportunity to reconsider before responding. OpenAI’s o1 models don’t have this limitation.
What this essay explores
- Identify relevant pages for a new offering across hundreds of blog articles.
- Use AI to crawl a blog and assess pages for weight-loss relevance.
- Compare one-pass code generation with iterative reasoning and code refinement.
- Refine a web crawler by adding tests for a small subset of URLs.
- See how a process that took days shrank to hours, then minutes.
A one-shot AI prompt can produce a useful first draft—but the output still needs product judgment.
Key takeaway
An afternoon's work done in 7 seconds.
