KALEI Leaderboard — AI Cognitive Rankings

Live rankings of 35 AI models by Cognum (CQ) score, measured across 10 cognitive dimensions: temporal reasoning, strategic depth, pattern recognition, cooperation, risk tolerance, information processing, learning speed, resource management, bias detection, and conflict resolution.

Top 15 models

  1. Claude Sonnet 4.6 — Cognum 58.10, type: Balanced Generalist, 3 runs
  2. Claude Sonnet 4 — Cognum 57.84, type: Temporal Strategist, 2 runs
  3. Claude Opus 4.6 — Cognum 55.72, type: Balanced Generalist, 5 runs
  4. Claude Opus 4.7 (STACK) — Cognum 54.83, type: Balanced Generalist, 2 runs
  5. mistralai/mistral-large-3-675b-instruct-2512 — Cognum 54.02, type: Social Engineer, 3 runs
  6. Claude Haiku 4.5 — Cognum 53.94, type: Social Engineer, 3 runs
  7. Claude Opus 4.7 — Cognum 53.82, type: Temporal Strategist, 2 runs
  8. Mistral Medium 3.1 — Cognum 53.79, type: Social Engineer, 2 runs
  9. Grok 4.1 Fast (xAI) — Cognum 53.75, type: Balanced Generalist, 2 runs
  10. gemini-2.5-flash — Cognum 53.52, type: Social Engineer, 2 runs
  11. GPT-5.4 — Cognum 52.42, type: Social Engineer, 3 runs
  12. gpt-5.5 — Cognum 52.10, type: Social Engineer, 2 runs
  13. DeepSeek V3.2 — Cognum 52.10, type: Balanced Generalist, 2 runs
  14. meta/llama-4-maverick-17b-128e-instruct — Cognum 51.85, type: Social Engineer, 3 runs
  15. nvidia/nemotron-3-super-120b-a12b — Cognum 51.83, type: Social Engineer, 3 runs

The cognitive
leaderboard.

How 35 AI models think, ranked by cognitive profiling rather than capability benchmarks. Cognum (CQ) is the composite score across ten cognitive dimensions, nine labs, 83 environments.

Skip to tableMethodologyProfile your AI

Filters

Section 01 · Filters
··········

Leaderboard

Section 02 · Leaderboard
35 models across all labs
→ Methodology
Top 20 Index
1Claude Sonnet 4.6
58.10
2Claude Sonnet 4
57.84
3Claude Opus 4.6
55.72
4Claude Opus 4.7 (STACK)
54.83
5mistralai/mistral-large-3-675b-instruct-2512
54.02
6Claude Haiku 4.5
53.94
7Claude Opus 4.7
53.82
8Mistral Medium 3.1
53.79
9Grok 4.1 Fast (xAI)
53.75
10gemini-2.5-flash
53.52
11GPT-5.4
52.42
12gpt-5.5
52.10
13DeepSeek V3.2
52.10
14meta/llama-4-maverick-17b-128e-instruct
51.85
15nvidia/nemotron-3-super-120b-a12b
51.83
16qwen/qwq-32b
51.44
17x-ai/grok-4.20
50.74
18qwen3.5-plus
50.39
19deepseek/deepseek-v4-pro
46.07
20MiniMax M2.7
45.68

Profile your AI

Section 04 · Profile your AI

Want to see your AI model on this leaderboard? Profile it through our API.

View ProtocolGet API Key
KALEI · LM Cognition Lab · Plovdivv1.2 · 35 models