BenchLeader
Menu
Best LLMsChoose a modelCoding plansLeaderboardsModelsBenchmarksProvidersReleases7

GDPval-AA v2.1, normalised score leaderboard

As of 6 Nov 2026, the top score on GDPval-AA v2.1, normalised score is 68.3% by Claude Opus 5.5 (Anthropic), ahead of Muse Wonder Full at 67.8%, across 167 model configurations with a published result (Artificial Analysis, result dated 22 Sep 2026). GDPval asks models to produce real work products across 44 occupations and has professionals compare them with a human expert's. Artificial Analysis runs the public set and normalised the win rate. GDPval-AA v2.1, normalised score is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 6 Nov 2026)

  • Claude Opus 5.5 holds the top GDPval-AA v2.1, normalised score score at 68.3% (Artificial Analysis, 22 Sep 2026).
  • Muse Wonder Full is second on GDPval-AA v2.1, normalised score at 67.8% (Artificial Analysis, 12 Oct 2026).
  • GDPval-AA v2.1, normalised score: Artificial Analysis' normalised run of OpenAI's GDPval professional-task comparison, frozen at the last percentages it published (25 October 2026). AA now reports the same evaluation as an Elo (Artificial Analysis).
  • 167 model configurations have a published GDPval-AA v2.1, normalised score result on BenchLeader (BenchLeader, 6 Nov 2026).
  • GDPval-AA v2.1, normalised score does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Artificial Analysis' normalised run of OpenAI's GDPval professional-task comparison, frozen at the last percentages it published (25 October 2026). AA now reports the same evaluation as an Elo.

Published by
Artificial Analysis
Category
Agents & tools
Index weight
Reference only
Models
167
Data as of
6 Nov 2026

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

This board is frozen. A different, newer test now carries its name, and its scores are on GDPval-AA v2.1.

What the test looks like

GDPval asks models to produce real work products across 44 occupations and has professionals compare them with a human expert's. Artificial Analysis runs the public set and normalised the win rate.

How it is scored

Normalised win rate against the human deliverable, frozen at the last percentages Artificial Analysis published, on 25 October 2026.

What to keep in mind

Artificial Analysis now reports this evaluation as an Elo, which is a different scale; those ratings are on the GDPval-AA board. The Epoch-mirrored official results are counted in the index.

Top 15
  1. 1Claude Opus 5.5 (max)68.3%
  2. 2Muse Wonder Full (max)67.8%
  3. 3Kimi K3.1 Pro67.3%
  4. 4Claude Sonnet 5.5 (max)66.9%
  5. 5Claude Opus 5.5 (xhigh)66.5%
  6. 6Muse Wonder (max)63.5%
  7. 7Claude Fable 5.1 (max)62.7%
  8. 8DeepSeek V4.1 Flash 1017 (max)62.3%
  9. 9Claude Fable 5.1 (xhigh)61.6%
  10. 10Claude Sonnet 5.5 (xhigh)61.4%
  11. 11Grok 4.7 (xhigh)60.7%
  12. 12Grok 4.7 (high)60.4%
  13. 13Claude Opus 5.5 (high)60.2%
  14. 14MiMo-V2.6-Pro59.1%
  15. 15Qwen3.8 Max (max)58.3%
167 of 167
#
1Claude Opus 5.5max68.3%69.32026-09-22
2Muse Wonder Fullmax67.8%62.92026-10-12
3Kimi K3.1 Pro67.3%61.72026-10-23
4Claude Sonnet 5.5max66.9%65.42026-09-28
5Claude Opus 5.5xhigh66.5%66.32026-09-22
6Muse Wondermax63.5%62.32026-10-12
7Claude Fable 5.1max62.7%67.22026-09-01
8DeepSeek V4.1 Flash 1017max62.3%59.22026-10-17
9Claude Fable 5.1xhigh61.6%65.42026-09-01
10Claude Sonnet 5.5xhigh61.4%63.42026-09-28
11Grok 4.7xhigh60.7%58.32026-09-21
12Grok 4.7high60.4%57.72026-09-21
13Claude Opus 5.5high60.2%65.82026-09-22
14MiMo-V2.6-Pro59.1%57.32026-09-21
15Qwen3.8 Maxmax58.3%58.82026-09-02
16GLM 5.3max57.5%59.02026-08-18
17Ling 3.1 Flash57.3%51.22026-10-22
18GLM 5.3 Flash57.1%56.82026-08-26
19Claude Fable 5.1high56.8%64.02026-09-01
20Qwen3.8-Flash-Next56.4%54.62026-08-26
21Gemini 4 Argonhigh56.2%66.02026-09-30
22Claude Haiku 5.5max56.0%57.02026-10-07
23MiMo-V2.6-Flash55.4%55.02026-09-21
24Qwen3.8 2.4T A95B55.4%52.02026-08-12
25GPT-6.1 Solmax54.8%67.32026-09-29
26Step 5 Preview54.3%54.92026-09-18
27Claude Opus 5.5medium54.3%62.02026-09-22
28Grok 4.7low54.1%56.02026-09-21
29GPT-6 Astramax53.7%67.82026-09-03
30GPT-6 Astraxhigh52.9%66.82026-09-03
31Claude Sonnet 5.5high52.4%58.62026-09-28
32Deepseek v4 Flash Visionmax52.3%–2026-08-21
33Claude Fable 5.1medium52.2%62.32026-09-01
34GPT-6.1 Solxhigh52.0%64.62026-09-29
35GPT-6 Astrahigh51.2%67.02026-09-03
36Claude Haiku 5.5xhigh50.7%57.32026-10-07
37GPT-6.1 Solhigh50.6%64.32026-09-29
38GPT-6 Astramedium49.3%65.12026-09-03
39Claude Fable 5.1low48.4%59.32026-09-01
40MiniMax M3.1 Flash47.9%53.02026-10-19
41GPT-5.6 Terramax47.6%60.72026-07-09
42GPT-6.1 Solmedium47.5%63.62026-09-29
43DeepSeek V4 Promax47.5%57.02026-08-13
44GPT-5.6 Terraxhigh47.4%60.32026-07-09
45Gemini 3.8 Flashhigh46.8%62.02026-09-02
46GPT-6 Lunamax46.6%56.72026-09-22
47Gemini 3.8 Flashmedium46.1%60.72026-09-02
48Mistral Large 446.1%52.92026-10-06
49Qwen3.8 27Bxhigh46.0%52.52026-08-14
50Claude Haiku 5.5high45.8%55.82026-10-07
51Muse Spark Openxhigh45.1%54.02026-10-15
52GPT-6 Astralow44.5%62.72026-09-03
53Qwen3.8 27Bmedium44.0%50.32026-08-14
54GPT-5.6 Terrahigh43.9%57.02026-07-09
55GPT-6 Lunaxhigh43.2%52.42026-09-22
56Qwen3.8 27Blow43.2%49.12026-08-14
57K2 Horizon 375B A23B42.9%49.82026-09-03
58GPT-6 Lunahigh42.1%51.42026-09-22
59Agnes 2.5 Pro Beta41.5%–2026-08-26
60GLM 5.3low41.2%47.62026-08-18

More agents & tools benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on GDPval-AA v2.1, normalised score

Cite as: BenchLeader, “GDPval-AA v2.1, normalised score leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/aa_gdpval.

GDPval-AA v2.1, normalised score: questions

What does GDPval-AA v2.1, normalised score measure?
GDPval asks models to produce real work products across 44 occupations and has professionals compare them with a human expert's. Artificial Analysis runs the public set and normalised the win rate. Scores are reported in percent of tasks solved; higher is better.
What is the top score on GDPval-AA v2.1, normalised score?
As of 6 Nov 2026, the highest score on GDPval-AA v2.1, normalised score is 68.3% by Claude Opus 5.5 (result dated 22 Sep 2026), ahead of Muse Wonder Full at 67.8%. Scores come from Artificial Analysis and are re-read daily.
Which AI model leads GDPval-AA v2.1, normalised score?
Claude Opus 5.5 leads GDPval-AA v2.1, normalised score with 68.3% as of 6 Nov 2026 (result dated 22 Sep 2026), ahead of Muse Wonder Full at 67.8%.
How many models have GDPval-AA v2.1, normalised score results?
167 model configurations have a GDPval-AA v2.1, normalised score result on BenchLeader, all taken from Artificial Analysis.
Who runs GDPval-AA v2.1, normalised score and how often is it updated?
GDPval-AA v2.1, normalised score is published by Artificial Analysis. BenchLeader re-reads the published results once a day and records the date each result was published.
Does GDPval-AA v2.1, normalised score count toward the BenchLeader Index?
No. GDPval-AA v2.1, normalised score is shown for reference but left out of the composite index.