Apodex TRACES — GPT-6 vs GPT-5.6 agent benchmark

A generational leap, but not on every dimension. Yesterday we put the TRACES board up. Here is the first comparison worth pulling out of it. GPT-6-astra @OpenAI tops the board but that is not the interesting part. Against GPT-5.6-sol, same harness, 140 agent episodes, it gains on all six capabilities. The gains are not even, and the uneven part is the story: ➡️ Tools: 2.34 → 2.80 ➡️ Repair: 1.81 → 2.01 ➡️ Alternatives: 2.10 → 2.80 ➡️ Coherence: 2.63 → 3.15 ➡️ Evidence: 2.26 → 2.76 ➡️ Scope: 2.20 → 2.75 Five of the six moved into the top three on the board. Repair did not. A single score tells you who won. TRACES tells you why, and where the frontier still breaks. Live leaderboard: View more