MMLU: Chemistry (subcategory)

The chemistry subcategory of MMLU: a rollup of the College Chemistry and High School Chemistry subjects, used by publishers that report MMLU at a coarser grain than all 57 subjects.

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categoryknowledge
Subcategorychemistry
Page statusactive
Metricaccuracy
Directionhigher_is_better
Unit%
Dataset size303
Dataset licenceMIT
PublisherUC Berkeley (original); Center for AI Safety (current host)

What it measures

This id does not correspond to a single dataset config in the Hugging Face mirror of MMLU. It corresponds to the "chemistry" subcategory the benchmark's authors define in the repository's categories.py, which pools the College Chemistry and High School Chemistry subjects -- general and organic chemistry at an undergraduate level, and standard high-school chemistry respectively. Some publishers report MMLU broken down by this kind of subcategory rather than by all 57 individual subjects; this id captures scores reported at that grain.

Task format

Four-option multiple-choice questions pooled from two underlying MMLU subjects (College Chemistry, High School Chemistry), graded on the single correct labelled option. How a given publisher averages the two subjects into one number -- an unweighted mean of per-subject accuracy, or a single accuracy over the pooled question set -- is not documented and not established here.

Models reporting this benchmark

These figures come from the model cards, which carry one collection date per card and no per-score attribution. They are shown as reported, not as verified evidence.
ModelProviderScoreCard as of
Claude Opus 4Anthropic82.52026-04
Claude Opus 4.6Anthropic82.52026-04
GPT-4.1OpenAI81.82026-04
Gemini 2.5 ProGoogle DeepMind81.52026-04
GPT-4oOpenAI80.22026-04
GPT-4o (2024-05-13)OpenAI80.22026-04
GPT-4o (2024-08-06)OpenAI80.22026-04
GPT-4o (2024-11-20)OpenAI80.22026-04
GPT-4o miniOpenAI80.22026-04
Claude Sonnet 4Anthropic79.82026-04
Claude Sonnet 4.5Anthropic79.82026-04
Claude Sonnet 4.5 (latest)Anthropic79.82026-04
DeepSeek R1DeepSeek78.52026-04
DeepSeek R1 0528DeepSeek78.52026-04
DeepSeek R1 0528 NVFP4 v2NVIDIA78.52026-04
DeepSeek R1 0528 Qwen3 8BDeepSeek78.52026-04
DeepSeek R1 Distill Llama 70BDeepSeek78.52026-04
DeepSeek R1 Distill Llama 8BDeepSeek78.52026-04
DeepSeek R1 Distill Qwen 1.5BDeepSeek78.52026-04
DeepSeek R1 Distill Qwen 14BDeepSeek78.52026-04
DeepSeek R1 Distill Qwen 32BDeepSeek78.52026-04
DeepSeek R1 Distill Qwen 7BDeepSeek78.52026-04
DeepSeek ReasonerDeepSeek78.52026-04
Qwen 3 235B InstructCerebras77.22026-04
Qwen3 235B-A22BAlibaba / Qwen Team77.22026-04
Mistral Large (latest)Mistral AI74.82026-04
Mistral Large 2.1Mistral AI74.82026-04
Mistral Large 3Mistral AI74.82026-04
Gemma 4 31BGoogle DeepMind74.52026-04
gemma 4 31B itGoogle DeepMind74.52026-04
gemma 4 31B it GGUFUnsloth74.52026-04
Gemma 4 31B IT NVFP4NVIDIA74.52026-04
Gemma 4 26BGoogle DeepMind73.22026-04
Llama 3.3 70B Instruct NVFP4NVIDIA72.52026-04
Llama-3.3-70B-InstructMeta72.52026-04
Llama 3.1 70BMeta71.22026-04
Llama 3.1 70B InstructMeta71.22026-04
phi 4Microsoft70.52026-04
Phi 4 mini instructMicrosoft70.52026-04
Phi 4 multimodal instructMicrosoft70.52026-04

Data

This page as JSON · Edit on GitHub