BenchLeader
Menu
Best LLMsChoose a modelLeaderboardsModelsBenchmarksProvidersReleases8

AA Intelligence Index v5.0

As of 29 Oct 2026, the top score on AA Intelligence Index v5.0 is 59.4 by Gemini 4 Carbon (Google), ahead of Claude Opus 5.5 at 58.4, across 80 model configurations with a published result (Artificial Analysis, result dated 28 Oct 2026). Artificial Analysis' Intelligence Index v5.0: nine evaluations equally weighted across Knowledge Work (AA-Briefcase, AutomationBench-AA, GDP.pdf), Coding (FrontierCode 2.0, Terminal-Bench 4.0), Science (Terminal-Bench-Science, CritPt v2 hard subset) and General (AA-Omniscience v1.1, HLE-Diamond). AA Intelligence Index v5.0 is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 29 Oct 2026)

  • Gemini 4 Carbon holds the top AA Intelligence Index v5.0 score at 59.4 (Artificial Analysis, 28 Oct 2026).
  • Claude Opus 5.5 is second on AA Intelligence Index v5.0 at 58.4 (Artificial Analysis, 22 Sept 2026).
  • AA Intelligence Index v5.0: Artificial Analysis' composite across Knowledge Work, Coding, Science and General (25% each): AA-Briefcase, AutomationBench-AA, GDP.pdf, FrontierCode 2.0, Terminal-Bench 4.0, Terminal-Bench-Science, CritPt v2 hard subset, AA-Omniscience v1.1 and HLE-Diamond (Artificial Analysis).
  • 80 model configurations have a published AA Intelligence Index v5.0 result on BenchLeader (BenchLeader, 29 Oct 2026).
  • AA Intelligence Index v5.0 does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Artificial Analysis' composite across Knowledge Work, Coding, Science and General (25% each): AA-Briefcase, AutomationBench-AA, GDP.pdf, FrontierCode 2.0, Terminal-Bench 4.0, Terminal-Bench-Science, CritPt v2 hard subset, AA-Omniscience v1.1 and HLE-Diamond.

Published by
Artificial Analysis
Category
Composite
Index weight
Reference only
Models
80
Data as of
29 Oct 2026
Newest result
29 Oct 2026

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

Not comparable with the earlier test of the same name, whose final scores stay on AA Intelligence Index v4.3.2.

What the test looks like

Artificial Analysis' Intelligence Index v5.0: nine evaluations equally weighted across Knowledge Work (AA-Briefcase, AutomationBench-AA, GDP.pdf), Coding (FrontierCode 2.0, Terminal-Bench 4.0), Science (Terminal-Bench-Science, CritPt v2 hard subset) and General (AA-Omniscience v1.1, HLE-Diamond).

How it is scored

Equal weight across the four categories (25% each); within a category each evaluation is equal. Reported as a 0–100 index. About 60% of the score comes from private or held-out evaluations.

What to keep in mind

Its membership changes between versions, which shifts every score at once. Index v5 dropped GDPval-AA, SciCode and AA-LCR from the composite; those boards are still published separately. It is one publisher's view of what matters; the BenchLeader Index draws on many sources, including this composite and several of its components.

Top 15
  1. 1Gemini 4 Carbon (high)59.4
  2. 2Claude Opus 5.5 (max)58.4
  3. 3Claude Opus 5.5 (xhigh)57.0
  4. 4GPT-6 Astra (max)56.9
  5. 5GPT-6 Astra (xhigh)55.3
  6. 6GPT-6 Astra (high)54.9
  7. 7GPT-6.1 Sol (max)54.5
  8. 8Claude Sonnet 5.5 (max)54.2
  9. 9GPT-6.1 Sol (xhigh)52.3
  10. 10Claude Opus 5.5 (high)51.7
  11. 11GPT-6 Astra (medium)51.5
  12. 12GPT-6.1 Terra (max)51.0
  13. 13Gemini 4 Argon (high)50.4
  14. 14GPT-6.1 Sol (high)50.2
  15. 15Muse Wonder Full (max)49.7
80 of 80
#
1NewGemini 4 Carbonhigh59.465.32026-10-28
2Claude Opus 5.5max58.469.22026-09-22
3Claude Opus 5.5xhigh57.066.02026-09-22
4GPT-6 Astramax56.967.72026-09-03
5GPT-6 Astraxhigh55.366.62026-09-03
6GPT-6 Astrahigh54.966.82026-09-03
7GPT-6.1 Solmax54.567.22026-09-29
8Claude Sonnet 5.5max54.265.02026-09-28
9GPT-6.1 Solxhigh52.364.32026-09-29
10Claude Opus 5.5high51.765.62026-09-22
11GPT-6 Astramedium51.564.92026-09-03
12NewGPT-6.1 Terramax51.064.92026-10-26
13Gemini 4 Argonhigh50.465.92026-09-30
14GPT-6.1 Solhigh50.264.02026-09-29
15Muse Wonder Fullmax49.762.22026-10-12
16Claude Fable 5.1xhigh49.764.92026-09-01
17Claude Sonnet 5.5xhigh49.663.22026-09-28
18Claude Fable 5.1max49.067.02026-09-01
19Claude Opus 5.5medium48.161.72026-09-22
20Claude Fable 5.1high47.063.62026-09-01
21GPT-6.1 Solmedium46.163.32026-09-29
22NewKimi K3.1 Pro4661.32026-10-23
23NewGPT-6.1 Terraxhigh45.060.32026-10-26
24Muse Wondermax43.362.22026-10-12
25GPT-6 Astralow43.062.32026-09-03
26Claude Fable 5.1medium43.062.02026-09-01
27NewGPT-6.1 Terrahigh42.360.02026-10-26
28Claude Sonnet 5.5high41.158.12026-09-28
29Claude Fable 5.1low38.558.82026-09-01
30NewGPT-6.1 Terramedium38.158.82026-10-26
31Muse Spark 1.3max36.461.82026-09-02
32Claude Haiku 5.5max36.256.72026-10-07
33Grok 4.7high35.057.02026-09-21
34GPT-6.1 Sollow34.560.32026-09-29
35Claude Opus 5.5low34.559.02026-09-22
36Grok 4.7xhigh34.358.12026-09-21
37Claude Haiku 5.5xhigh34.356.82026-10-07
38Qwen3.8 Maxmax33.558.82026-09-02
39DeepSeek V4.1 Flash 1017max32.559.32026-10-17
40GLM 5.3max31.658.92026-08-18
41Grok 4.7low30.955.42026-09-21
42Gemini 3.8 Flashhigh30.661.92026-09-02
43NewGPT-6.1 Terrano reasoning30.451.82026-10-26
44Claude Sonnet 5.5medium30.350.42026-09-28
45Gemini 3.8 Flashmedium30.060.32026-09-02
46MiMo-V2.6-Pro29.957.22026-09-21
47Claude Haiku 5.5high29.655.52026-10-07
48Ling 3.1 Flash29.550.32026-10-22
49NewGPT-6.1 Terralow29.256.72026-10-26
50MiniMax M3.1 Flash28.952.92026-10-19
51Step 5 Preview28.554.82026-09-18
52GPT-6 Lunamax27.856.62026-09-22
53MiMo-V2.6-Flash26.755.02026-09-21
54Mistral Large 4high26.049.02026-10-29
55Claude Sonnet 5.5low25.350.42026-09-28
56Mistral Large 425.350.52026-10-06
57Muse Spark Openxhigh24.753.22026-10-15
58Claude Haiku 5.5medium24.649.82026-10-07
59GPT-6 Lunaxhigh23.351.82026-09-22
60Mistral Large 4medium23.1–2026-10-29

More composite benchmarks · All LLM benchmarks

Related benchmarks

Top head-to-heads on AA Intelligence Index v5.0

Cite as: BenchLeader, “AA Intelligence Index v5.0 leaderboard”, data as of 29 Oct 2026, https://www.benchleader.com/benchmarks/aa_intelligence_index_v5.

AA Intelligence Index v5.0: questions

What does AA Intelligence Index v5.0 measure?
Artificial Analysis' Intelligence Index v5.0: nine evaluations equally weighted across Knowledge Work (AA-Briefcase, AutomationBench-AA, GDP.pdf), Coding (FrontierCode 2.0, Terminal-Bench 4.0), Science (Terminal-Bench-Science, CritPt v2 hard subset) and General (AA-Omniscience v1.1, HLE-Diamond). Scores are reported in index; higher is better.
What is the top score on AA Intelligence Index v5.0?
As of 29 Oct 2026, the highest score on AA Intelligence Index v5.0 is 59.4 by Gemini 4 Carbon (result dated 28 Oct 2026), ahead of Claude Opus 5.5 at 58.4. Scores come from Artificial Analysis and are re-read daily.
Which AI model leads AA Intelligence Index v5.0?
Gemini 4 Carbon leads AA Intelligence Index v5.0 with 59.4 as of 29 Oct 2026, ahead of Claude Opus 5.5 at 58.4 (result dated 28 Oct 2026).
How many models have AA Intelligence Index v5.0 results?
80 model configurations have a AA Intelligence Index v5.0 result on BenchLeader, all taken from Artificial Analysis.
Who runs AA Intelligence Index v5.0 and how often is it updated?
AA Intelligence Index v5.0 is published by Artificial Analysis. BenchLeader re-reads the published results once a day and records the date each result was published.
Does AA Intelligence Index v5.0 count toward the BenchLeader Index?
No. AA Intelligence Index v5.0 is shown for reference but left out of the composite index.