BenchLeader
Menu
LeaderboardsModelsProvidersBenchmarksCompareReleases8BlogMethodology

AA-Briefcase

Knowledge work delivered as real files — spreadsheets, decks, documents — graded head to head, so the score is an Elo rather than a percentage.

As of 22 Sept 2026, Claude Opus 5.5 leads AA-Briefcase on BenchLeader with 1822, ahead of Claude Opus 5.5 at 1780, across 51 model configurations with a published result.

Published by
Artificial Analysis
Category
Agents & tools
Index weight
Reference only
Models
51
Data as of
22 Sept 2026

Source: Artificial Analysis (artificialanalysis.ai). Data taken from the public leaderboard.

Top 15
  1. 1Claude Opus 5.5 (thinking)1822
  2. 2Claude Opus 5.5 (xhigh)1780
  3. 3Claude Opus 5.5 (high)1705
  4. 4Claude Fable 5.1 (thinking)1678
  5. 5Claude Opus 5 (max)1673
  6. 6Claude Fable 5.1 (xhigh)1669
  7. 7Grok 4.7 (xhigh)1657
  8. 8Claude Opus 5 (xhigh)1649
  9. 9Grok 4.7 (high)1644
  10. 10Qwen3.8 Max (0902) (max)1640
  11. 11Muse Spark 1.3 (max)1597
  12. 12Claude Fable 5.1 (high)1592
  13. 13Claude Opus 5 (high)1573
  14. 14GPT-6 Astra (max)1569
  15. 15Grok 4.6 (xhigh)1555
51 of 51
#
1NewClaude Opus 5.5thinkingAnthropic182270.72026-09-22
2NewClaude Opus 5.5xhighAnthropic178070.32026-09-17
3NewClaude Opus 5.5highAnthropic170569.92026-09-17
4Claude Fable 5.1thinkingAnthropic167871.02026-09-01
5Claude Opus 5maxAnthropic167369.52026-07-24
6Claude Fable 5.1xhighAnthropic166971.12026-09-01
7NewGrok 4.7xhighSpaceXAI165761.92026-09-21
8Claude Opus 5xhighAnthropic164969.32026-07-24
9NewGrok 4.7highSpaceXAI1644–2026-09-21
10Qwen3.8 Max (0902)maxAlibaba164066.22026-09-02
11Muse Spark 1.3maxMeta159769.22026-09-02
12Claude Fable 5.1highAnthropic159271.02026-09-01
13Claude Opus 5highAnthropic157369.82026-07-24
14GPT-6 AstramaxOpenAI156971.72026-09-03
15Grok 4.6xhighSpaceXAI155564.12026-08-12
16Grok 4.6highSpaceXAI154664.22026-08-12
17GPT-6 AstraxhighOpenAI154470.52026-09-03
18Claude Fable 5thinkingAnthropic154370.32026-06-09
19Claude Fable 5.1mediumAnthropic154269.12026-09-01
20GLM 5.3maxZhipu AIopen ↗152565.42026-08-18
21NewMiMo-V2.6-ProXiaomiopen1522–2026-09-21
22Kimi K3maxMoonshot AIopen ↗151066.72026-07-16
23GPT-6 AstrahighOpenAI150771.32026-09-03
24Muse Spark 1.3xhighMeta149467.92026-09-02
25GPT-5.6 SolmaxOpenAI148768.52026-07-09
26GLM 5.3 FlashZhipu AIopen ↗145963.62026-08-26
27GPT-6 AstramediumOpenAI145969.22026-09-03
28Qwen3.8 2.4T A95BAlibabaopen ↗144564.52026-08-12
29Claude Opus 5mediumAnthropic144166.42026-07-24
30NewStep 5 PreviewStepFun143264.82026-09-18
31DeepSeek V4.1 FlashmaxDeepSeekopen ↗143261.42026-09-10
32Qwen3.8 27BxhighAlibabaopen ↗140358.92026-08-14
33Claude Sonnet 5maxAnthropic136360.02026-06-30
34GPT-5.6 LunamaxOpenAI134559.92026-07-09
35Muse Spark 1.2xhighMeta134063.72026-08-05
36GPT-5.6 TerramaxOpenAI133664.82026-07-09
37K2 Horizon 375B A23BMBZUAIopen130357.22026-09-03
38DeepSeek V4 PromaxDeepSeekopen ↗126163.72026-08-13
39GPT-6 AstralowOpenAI126167.62026-09-03
40Gemini 3.8 FlashhighGoogle120264.32026-09-02
41MiniMax M3MiniMaxopen ↗109257.22026-06-01
42Nemotron 3 Ultra 550B A55BthinkingNVIDIAopen ↗87357.32026-06-04
43Gemini 3.5 FlashhighGoogle86863.42026-05-19
44Gemini 3.5 Flash LiteGoogle64255.22026-07-21
45Mistral Medium 3.5Mistral AIopen51551.12026-04-29
46Nemotron 3.5 LightningNVIDIAopen ↗50048.02026-08-11
47Muse GlimmerhighMetaopen ↗46952.52026-08-10
48Granite 4.2 8BIBMopen ↗30246.72026-08-25
49Granite 4.2 3BIBMopen ↗9445.52026-08-25
50Nemotron 3 Nano 30B A3BthinkingNVIDIAopen ↗7547.02025-12-15
51K2 Think V2thinkingMBZUAIopen448.42025-12-15

Cite as: BenchLeader, “AA-Briefcase leaderboard”, https://www.benchleader.com/benchmarks/aa_briefcase, data as of 22 Sept 2026.

AA-Briefcase: questions

What does AA-Briefcase measure?
Knowledge work delivered as real files — spreadsheets, decks, documents — graded head to head, so the score is an Elo rather than a percentage. Scores are reported in Elo-style ratings from pairwise comparisons; higher is better.
Which AI model leads AA-Briefcase?
Claude Opus 5.5 leads AA-Briefcase with 1822 as of 22 Sept 2026, ahead of Claude Opus 5.5 at 1780.
How many models have AA-Briefcase results?
51 model configurations have a AA-Briefcase result on BenchLeader, all taken from Artificial Analysis.
Who runs AA-Briefcase and how often is it updated?
AA-Briefcase is published by Artificial Analysis. BenchLeader re-reads the published results every morning and records the date each result was published.
Does AA-Briefcase count toward the BenchLeader Index?
No. AA-Briefcase is shown for reference but left out of the composite index.