Argo-Bench - Solved: leaderboard

Argo-Bench (TextQL, 2026) solved rate — share of the 210 enterprise data-science tasks on a simulated Oracle EBS warehouse on which an agent scores 95 or more out of 100, one row per reasoning effort.

Metric: Share of the 210 tasks solved (%), a task counting as solved at a score of 95 or more (v1.1); one row per reasoning effort. Source: argo-bench.com. Saturation forecast: Around March 2028. 21 models tracked. Updated 16 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Fable 5.2xHigh41.43#1
2Gemini 4 CarbonHigh38.1#2
3Gemini 4 ArgonHigh33.81#12
4Claude Opus 5.5xHigh33.33#5
5Claude Sonnet 5.5xHigh26.19#19
6GPT-6 AstraxHigh25.71#6
7GPT-6.1 TerraHigh24.29#21
8GPT-6.1 SolxHigh22.38#10
9Muse WonderHigh21.9#28
10Grok 4.7High20.48#47
11DeepSeek V4.1 FlashMax15.71#62
12GPT-6 SolxHigh15.71#26
13Gemini 3.8 FlashHigh14.76#36
14DeepSeek V4.1 Flash (1017)xHigh13.81#41
15Kimi K3xHigh12.38#56

Interactive version: theaggregate.ai/benchmark?slug=argo-bench-solved · How It Works · Data refreshed daily, snapshot 2026-11-16.