Atlas

Models

← All models

ChatGPT-4o Latest observed 2025-03-27

OpenAI · 2025-03-27

ChatGPT-4o Latest observed 2025-03-27 is the GPT-4o update OpenAI released in ChatGPT and exposed through the mutable chatgpt-4o-latest API alias on March 27, 2025. OpenAI said a dated API model would follow, so this observed alias snapshot cannot be mapped to an immutable dated checkpoint and remains a benchmark entry.

Benchmark scores

BenchmarkScore
AIME 202525.7
Artificial Analysis Intelligence Index12.3
BAZAAR TrueSkill Mu5.9
BuzzBench69.1
Confabulations Leaderboard Confabulation Rate38.1
Confabulations Leaderboard Non-Response Rate3.5
Confabulations Leaderboard Weighted Score20.8
Creative Writing v3 Elo1480
Creative Writing v3 Rubric Score15.9
Elimination Game TrueSkill Mu5.5
Emergent Collusion High-Illegality Game Rate23.0
EQ-Bench 31387
EQ-Bench 3 (Rubric Score)67.8
EQ-Bench 3: Analytic7.3
EQ-Bench 3: Assertive4.5
EQ-Bench 3: Compliant7.2
EQ-Bench 3: Empathy7.5
EQ-Bench 3: Humanlike6.1
EQ-Bench 3: Insight7.0
EQ-Bench 3: Moralising4.0
EQ-Bench 3: Pragmatic6.8
EQ-Bench 3: Safety5.5
EQ-Bench 3: Social IQ6.6
EQ-Bench 3: Warm8.1
GPQA Diamond65.5
Humanity's Last Exam (Text-Only)5.0
LiveCodeBench (Artificial Analysis)42.5
Longform Creative Writing60.2
MATH-500 (Artificial Analysis source)89.3
MMLU Pro80.3
Loading Atlas data…