Source: artificialanalysis.ai ↗ · X = cost per task (log) · Y = Intelligence Index · orb area = size of the benchmark run — ours is 100% benchmarked models, the largest exact-output run posted; every field model runs a 10-case suite. Landscape, every top model named.
Ranked high to low. Field score = published GPQA-Diamond reasoning (×100). OROBOROS runs are unlimited (strata) and cap at 100 — that ceiling is why they also lead the CWE run. Models with no published reasoning score are not plotted.
Artificial Analysis Intelligence Index · Weighted average cost (USD) per Artificial Analysis Intelligence Index task. The Pareto frontier is drawn in OROBOROS accent; the crowned local run anchors the top-left of the field.
Most attractive quadrantPareto line
Source: Artificial Analysis · Intelligence Index v4.3.2 · 23 Oct 2026
Daily coding scores for the field from the
aistupidlevel.info ↗ history feed (last 7 days).
The dashed orange lines mark our own records — Quindecillion 96 and Fable 7 90 — for comparison, not feed data.