Atlas

Models

← All models

ChatGPT-4o Latest (September 2024 benchmark entry)

OpenAI · 2024-08-15

ChatGPT-4o Latest (September 2024 benchmark entry) represents OpenAI's mutable `chatgpt-4o-latest` alias under benchmark source labels dated 2024-09-03 (EQ-Bench) and 2024-09-07 (ZeroEval). OpenAI launched the dynamic alias on 2024-08-15 and documented it as pointing to the latest GPT-4o model used by ChatGPT. Neither benchmark source identifies an immutable underlying API checkpoint, so this entry remains separate from dated GPT-4o artifacts.

Benchmark scores

BenchmarkScore
Creative Writing v282.5
EQ-Bench v282.5
EQ-Bench v2 + MAGI-Hard Combined81.5
MAGI-Hard80.6
ZeroEval64.6
ZeroEval CRUX86.5
ZeroEval MATH Level 553.1
ZeroEval MMLU-Redux88.9
ZeroEval ZebraLogic29.9
Loading Atlas data…