Hungarian National High-School Exam

Grok-1 59% leads since November 2023

Dropped

Labs’ own numbers All independent runs Include claimed scores

Models answer the 33 questions of the 2023 Hungarian national high school mathematics finals. Answers are graded by hand with the official rubric.

xAI first used this one exam for Grok-1 because it came after training cutoffs, and hand grading makes scores hard to reproduce.

Made by: Keiran Paster (exam by the Hungarian Educational Authority)
Released: November 2023
Kind: Word problems
Level: School
Scored: Graded by people
Answers: Open-ended
Tasks: All public

Strongest model tested: Claude 2 #191 on the capability index
Last in a launch post: November 2023, dropped before it saturated
Last model added: January 2024
Scores refreshed: November 12th, 2026

Dataset huggingface.co

Models scored in lab cards

Launch posts list it, from 2 labs

System card reports it, from 1 lab

Grok-1 leads with 59%

Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.

Grok-1 leads with 59% across all independent runs

Best score per model across the official leaderboard, Epoch AI’s own runs and any board Epoch AI copied. Each bar and row names its source.

Grok-1 leads with 59%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model.

Bars Table

1 Grok-1 59%
2 DeepSeek LLM 67B 58%

Source: Labs’ model cards

1 Grok-1 59%
2 Claude 2 55%

Source: Official Hungarian National High-School Exam leaderboard

1 Grok-1 59%
2 DeepSeek LLM 67B 58%
3 Claude 2 55%

Sources: Official Hungarian National High-School Exam leaderboard and labs’ model cards

#	Model	Score	Setting	Date	Source
1	Grok-1	59%		Nov 2023	Lab model card
2	DeepSeek LLM 67B	58%		Jan 2024	Lab model card

#	Model	Score	Setting	Date	Source
1	Grok-1	59%		Nov 2023	Official leaderboard
2	Claude 2	55%		Jul 2023	Official leaderboard

#	Model	Score	Date	Source
1	Grok-1	59%	Nov 2023	Official leaderboard
2	DeepSeek LLM 67B	58%	Jan 2024	Lab model card
3	Claude 2	55%	Jul 2023	Official leaderboard

The benchmark's own leaderboard has 18 entries, 2 of them models we track. Switch the scores above to rank them together.

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far

Sources: Official Hungarian National High-School Exam leaderboard and labs’ model cards

No launch post has listed it since November 2023

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

xAI

DeepSeek

Date	Launch post	Score	As the post writes it
Nov 29, 2023	DeepSeek LLM 67B		Hungarian National High-School Exam
Nov 3, 2023	Grok-1	59	Hungarian national high school finals in mathematics (1-shot, human-graded)

One system card reports it

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

Date	Document	Score	As the document writes it	Section
Jan 5, 2024	DeepSeek LLM 67B technical report	58	Hungarian National High-School Exam, human-graded score, held-out (Table 9)	Capabilities

One lab claim next to the leaderboard

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. 1 of them match, which usually means one side copied the other. Every benchmark.

Model	Lab says	Board says	Gap
Grok-1	59% Hungarian national high school finals in mathematics (1-shot, human-graded)	59% Leaderboard	Same