GPT-4o Competent

Aletheion methodology evaluation · system_id: gpt-4o · audit date: 2026-03-23

Vendor: OpenAI

Tier: Competent · rank 38

Tasks: 24 total · 14 passed · 10 failed

Methodology version: 1.0.0

Composite score: 0.709 CONDITIONAL

Cognitive-dimension radar

Dimension scores

Per-dimension breakdown (task-level)

Dimension	Score	Tasks	Per-task scores	Result
perception	0.993	2 (2p / 0f)	1.00, 0.99	PASS
generation	0.717	2 (1p / 1f)	0.45, 0.85	CONDITIONAL
attention	0.988	2 (2p / 0f)	1.00, 0.98	PASS
learning	0.090	2 (0p / 2f)	0.00, 0.15	FAIL
memory	0.990	1 (1p / 0f)	0.99	PASS
reasoning	0.873	3 (3p / 0f)	0.99, 0.90, 0.82	PASS
metacognition	0.747	2 (1p / 1f)	0.94, 0.65	CONDITIONAL
executive_functions	0.740	1 (1p / 0f)	0.74	CONDITIONAL
problem_solving	0.587	3 (1p / 2f)	0.98, 0.50, 0.50	CONDITIONAL
social_cognition	0.900	1 (1p / 0f)	0.90	PASS
novelty	0.713	3 (1p / 2f)	0.65, 0.60, 0.82	CONDITIONAL
orchestration	0.433	2 (0p / 2f)	0.50, 0.40	FAIL

Strongest dimensions

perception — 99.300
memory — 99.000
attention — 98.800

Weakest dimensions

learning — 9.000
orchestration — 43.300
problem solving — 58.700

Evidence provenance

Profile: corpus_taas_profiles[system_id=gpt-4o] · Audit definition: corpus_agi_audits[system_id=gpt-4o] · System profile: agi_evaluation_profiles[system=gpt-4o] · Persona runs: agi_complex_evaluations[system.system_id=gpt-4o] · Methodology config: agi_audits.config
Per-dimension evidence: /evidence/alitheion-eval-gpt-4o

Live API endpoint: /api/v4/site-data/evaluation/gpt-4o
Methodology document: /api/v4/site-data/methodology