Acropolis

Qwen3.7 Max

AlibabaGrade B. Evidence grade B of A to D: good.ProprietaryReleased 19 May 2026Within-domain variance
Compare with peers

Domain profile

Median within-board percentile by capability domain across boards with sufficient peer overlap.

  • Agentic (2 boards): 51 (scale 0 to 100)Agentic (2 boards)51
  • Coding (3 boards): 57 (scale 0 to 100)Coding (3 boards)57
  • Science (1 boards): 83 (scale 0 to 100)Science (1 boards)83

Benchmark appearances

Best verified run per board, with confidence intervals and Pareto frontier status.

BenchmarkRankScoreCostStatusObserved
SWE-Bench verified

Epoch AI Benchmarking Hub · Default

#7 / 35
77.3%[73.5% to 81%]UnknownStandard18 Jun 2026
GPQA diamond

Epoch AI Benchmarking Hub · Default

#44 / 338
90.9%[86.9% to 94.9%]UnknownStandard7 Aug 2026
LMArena Agent

LMArena · Overall · Default

#52 / 64
-0.1[-0.1 to -0.1]$0.28Standard15 Nov 2026
LMArena WebDev

LMArena · Webdev react · Default

#62 / 140
1,512[1,504 to 1,520]UnknownStandard15 Nov 2026
LMArena WebDev

LMArena · Overall · Default

#65 / 156
1,515[1,508 to 1,523]UnknownStandard15 Nov 2026
LMArena WebDev

LMArena · Webdev · Default

#65 / 156
1,515[1,508 to 1,523]UnknownStandard15 Nov 2026
LMArena WebDev

LMArena · Webdev HTML · Default

#66 / 155
1,513[1,494 to 1,533]UnknownStandard15 Nov 2026

Closest head-to-head rivals

Direct pairwise win rates for Qwen3.7 Max across shared benchmark boards.