<p>AI suddenly feels like it has crossed a threshold, and Yann Dubois, co-lead of the Post-training Frontiers team at OpenAI, joins Matt Turck to explain why. Yann’s team has led the post-training behind the company&#39;s reasoning models, including the recent GPT-5.5 release. In this conversation, we go inside the shift from raw model capability to useful, reliable systems: what changed with GPT-5.5, why reinforcement learning is moving beyond math and coding competitions into messy real-world work, how reasoning models like GPT-5.5 actually work, the difference between GPT-5.5 Thinking and GPT-5.5 Pro, why post-training has become one of the most important frontiers in AI, and why evals, model-as-judge, hallucinations, agentic workflows, GDPval, and continual learning are now central to the next phase of frontier models. Yann also shares why continual learning remains one of AI&#39;s biggest unsolved problems three years after ChatGPT, and where startups still have massive room to build as frontier models race ahead.</p><p><br></p><p>(00:00) - Cold open</p><p>(00:34) - Intro</p><p>(01:30) - Why recent AI progress feels like a step function</p><p>(04:13) - Model reliability &amp; the rollercoaster of shipping 5.5</p><p>(07:33) - How OpenAI structures vertical and horizontal teams</p><p>(09:49) - Improving model efficiency and test-time compute</p><p>(12:32) - Yann Dubois&#39; journey from Switzerland to OpenAI</p><p>(15:37) - Reasoning in 2026: Real-world utility vs verifiable rewards</p><p>(18:34) - GPT-5.5 Thinking vs Pro: Scaling test-time compute</p><p>(20:09) - How reasoning models become more efficient</p><p>(23:23) - Pre-training scaling and overcoming the data wall</p><p>(27:03) - Multimodal data, synthetic data, and embodied AI</p><p>(31:05) - Demystifying mid-training and post-training</p><p>(37:21) - Does RL create new capabilities in AI?</p><p>(38:53) - The challenges and frontier of scaling RL</p><p>(43:09) - Is building AI models a craft or a strict science?</p><p>(48:21) - How AI models generalize across different domains</p><p>(54:18) - How reinforcement learning cures AI hallucinations</p><p>(56:04) - Negative generalization and conflicting instructions</p><p>(58:05) - Can RL scale to law, medicine, and the broader economy?</p><p>(1:00:19) - The evaluation bottleneck and Model as a Judge</p><p>(1:04:21) - Continuous AI progress &amp; continual learning</p><p>(1:08:49) - Will foundation models eat the agent harness?</p><p>(1:11:23) - Why startups should focus on the last mile of AI</p>