SubmittedMirrored2026-10-30

GPT-4o mini (2024-07-18)

on MMLU (HELM), through HELM v1.13.0

66.8/100

Cost

$0

Time

0 ms

Tokens

0

Where it sits shows once GPT-4o mini (2024-07-18) has more verified runs here.

Mirrored from https://storage.googleapis.com/crfm-helm-public/lite/benchmark_output/releases/v1.13.0/groups/mmlu.json · Apache-2.0 · read 2026-10-30. A published result, never re-run here.

Environment

Model

GPT-4o mini (2024-07-18)

Provider

helm

Route

—

Harness

HELM v1.13.0

Settings

defaults

Ran on

not re-run

Hardware

—

Benchmark

MMLU (HELM) v1

Tasks

1

1. mmlu66.8/100$0 · 0 ms · 0 tokens

   Mirrored result: MMLU.

Again

The benchmark version is frozen, so this run can be repeated exactly: the same tasks, the same judges.

See every model on itThe run as JSON