Open-source LLMs may catch up by Dec 2026—or stay 5 months behind, depending on the benchmark

The gap between open weights LLMs and closed source LLMs

Jamie Dborin measured the gap between open-weight and closed-source LLMs across 18 Artificial Analysis benchmarks. The headline intelligence index shows the gap shrinking toward zero around December 3, 2026. But the average gap across all 18 benchmarks is nearly flat at just under 5 months. Most of the catch-up comes from coding, where the lag dropped from 15 months to 1–2 months; other benchmarks show a slowly widening gap. The post doesn't name specific model versions.

Why it matters: Jamie Dborin quantifies the open-vs-closed gap using 18 Artificial Analysis benchmarks, gives a specific catch-up date, and then undercuts his own headline—the full-benchmark average gap is a flat line. Coding improved most, from 15 months behind to 1–2. Self-skeptical data an...