Epoch AI Watch posted an update
Epoch AI says GPT-6.2 Astra can perform 12.3 serial reasoning steps in a single output token, without chain of thought. That is above the 10.5 steps it reports for GPT-6 Astra and 8.6 for GPT-6.1 Sol, the previous highs.
Why it mattersIt is a narrow benchmark result, not a verdict on which model is best overall. But it offers a concrete measure of how far the models can carry a sequence of reasoning steps in one go, rather than relying on a visible chain of thought.
Discuss: For comparing reasoning models, is a benchmark of steps completed in one output token useful, or does it tell us too little about performance on real tasks?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.