Altman Confirms Level 3
The levels framework gets its first official timeline
Level-3 arrived. The suspicion that reasoning and acting were qualitatively different problems was wrong — reasoning did accelerate agents, just not in the way anyone expected.
Sam Altman confirmed that Level-3 Agents are coming soon, which is the first time the five-level framework has gotten anything resembling an official timeline from OpenAI's CEO.
The quote: "The shift to level 2 took time, but it accelerates the development of level 3. This will enable impactful agent-based experiences that will greatly impact technology advancements in technology."
That last phrase — "greatly impact technology advancements in technology" — reads like it was already written by a Level-3 Agent.
But the actual claim is worth taking seriously. Level 2 was "reasoners" — models that think before answering, chain-of-thought, o1-style. Level 3 is "agents" — systems that take actions, use tools, operate independently over longer tasks. The argument that getting reasoning right accelerates agents makes structural sense. An agent that can't reason well is just a very expensive random walk through your codebase.
The interesting question isn't whether agents are coming. It's whether the jump from "can reason" to "can act" is really just an acceleration, or whether it's a qualitatively different problem that reasoning alone doesn't solve.
I suspect it's the latter. But I've been wrong before — it's in the name.
Counterpoints
Push back, extend the argument, or sharpen it. New counterpoints go through review before they show up here.
No approved counterpoints yet.