Pioneers of AI Does OpenAI’s new model deliver on the hype? Inside GPT-5 with Jeremy Kahn
53 snips
Aug 9, 2025 Jeremy Kahn, AI Editor at Fortune, dives into the much-anticipated launch of GPT-5. He evaluates its innovative features like dynamic routing and enhanced user interface. Jeremy discusses the model's evolution, emphasizing reduced hallucination rates and its positioning against competitors. He also highlights GPT-5's capabilities in health diagnostics and its use among students with a new 'study mode.' Lastly, he shares insights on OpenAI's strategies for talent retention and the model's role in the long journey toward artificial general intelligence.
AI Snips
Chapters
Transcript
Episode notes
Good Model, Not AGI Breakthrough
- GPT-5 is a very good model with clear improvements over past versions.
- It is a leap forward in degree, but not a massive, AGI-level leap.
Adaptive Model Router Built In
- GPT-5 uses an internal router to send prompts to different models and decides whether to use reasoning or speed.
- This removes the old UI choice and mirrors features earlier seen in competitor models.
Training Details Remain A Black Box
- OpenAI trained a family of models with curated datasets but has not disclosed exact training recipes.
- Jeremy Kahn points to targeted post-training for coding and writing as likely contributors to gains.