Pro Mode — raw scores, one benchmark under the lensAA-graded harness
The field · 200 measured
49.66788.667
200 models measured, most on the AA-graded harness. Kimi K3 tops the board at 88.667.
200 measured·16 new this week·Lifecycle Updated Oct 9, 2026Score basis
#◆Vendor$/M in$/M out
Leader — ignites whiteField — fill cools with scoreBars show 49.667–88.667 · ◆ solid = first-party or better · outlined = arm's-length
07Claude Opus 5.5 (Adaptive Reasoning, Medium Effort, Default Fallback)NEWAnthropic84.333$4.0/M$20.0/M$0.142
24Claude Sonnet 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)NEW+2 altsAnthropic82.667$2.0/M$10.0/M$0.073
84Claude Sonnet 5.5 (Adaptive Reasoning, Low Effort, Default Fallback)NEWAnthropic76$2.0/M$10.0/M$0.079
◆ solid = independently verified · aggregator attested · vendor attributed — outlined = cross-reference · source-attributed. Dimmed ◆ = confidence inferred. “+N alts” = alternate disclosures in the drawer with why-demoted. Prices per million tokens; $/point = input price ÷ score. Every row opens its model dossier.
Verification: 200 models scored · 0 independently verified · 200 aggregator-attributed · 0 with source disagreement. How these tiers are assigned