Comment by pmxi

That's a summarized and filtered view of the actual reasoning.

OpenAI and Anthropic guard the real reasoning closely. Users have never been able to see it and the API returns an encrypted blob instead of legible reasoning.