MMLU: Biology (subcategory)

The biology subcategory of MMLU: a rollup of the College Biology and High School Biology 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
Subcategorybiology
Page statusactive
Metricaccuracy
Directionhigher_is_better
Unit%
Dataset size454
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 "biology" subcategory the benchmark's authors define in the repository's categories.py, which pools the College Biology and High School Biology subjects -- molecular, cellular and organismal biology at an undergraduate level, and standard high-school biology (ecology, genetics, cell biology) 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 Biology, High School Biology), 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 4Anthropic88.22026-04
Claude Opus 4.6Anthropic88.22026-04
GPT-4.1OpenAI87.52026-04
Gemini 2.5 ProGoogle DeepMind87.22026-04
GPT-4oOpenAI86.82026-04
GPT-4o (2024-05-13)OpenAI86.82026-04
GPT-4o (2024-08-06)OpenAI86.82026-04
GPT-4o (2024-11-20)OpenAI86.82026-04
GPT-4o miniOpenAI86.82026-04
Claude Sonnet 4Anthropic85.52026-04
Claude Sonnet 4.5Anthropic85.52026-04
Claude Sonnet 4.5 (latest)Anthropic85.52026-04
DeepSeek R1DeepSeek84.22026-04
DeepSeek R1 0528DeepSeek84.22026-04
DeepSeek R1 0528 NVFP4 v2NVIDIA84.22026-04
DeepSeek R1 0528 Qwen3 8BDeepSeek84.22026-04
DeepSeek R1 Distill Llama 70BDeepSeek84.22026-04
DeepSeek R1 Distill Llama 8BDeepSeek84.22026-04
DeepSeek R1 Distill Qwen 1.5BDeepSeek84.22026-04
DeepSeek R1 Distill Qwen 14BDeepSeek84.22026-04
DeepSeek R1 Distill Qwen 32BDeepSeek84.22026-04
DeepSeek R1 Distill Qwen 7BDeepSeek84.22026-04
DeepSeek ReasonerDeepSeek84.22026-04
Qwen 3 235B InstructCerebras83.82026-04
Qwen3 235B-A22BAlibaba / Qwen Team83.82026-04
Gemma 4 31BGoogle DeepMind81.52026-04
gemma 4 31B itGoogle DeepMind81.52026-04
gemma 4 31B it GGUFUnsloth81.52026-04
Gemma 4 31B IT NVFP4NVIDIA81.52026-04
Mistral Large (latest)Mistral AI81.52026-04
Mistral Large 2.1Mistral AI81.52026-04
Mistral Large 3Mistral AI81.52026-04
Gemma 4 26BGoogle DeepMind80.82026-04
Llama 3.3 70B Instruct NVFP4NVIDIA80.22026-04
Llama-3.3-70B-InstructMeta80.22026-04
Llama 3.1 70BMeta79.52026-04
Llama 3.1 70B InstructMeta79.52026-04
phi 4Microsoft77.82026-04
Phi 4 mini instructMicrosoft77.82026-04
Phi 4 multimodal instructMicrosoft77.82026-04

Data

This page as JSON · Edit on GitHub