Benchmark Almanac

K12-PEBench

K12-PEM 69.4% tops its leaderboard InternVL2.5-78B-MPO 63.4% leads the models we track
Not in model cards

Models identify incorrect reasoning steps in solutions to school problems. The evaluator compares their decisions with human annotations for 3,033 examples.

Made by
Chong Li and colleagues
Released
June 2025
Kind
Other professions
Level
School
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Language
Chinese
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Last model added
January 2025
Scores refreshed
November 20th, 2026
9
Entries in its paper, 6 models we track

K12-PEM tops the leaderboard with 69.4% InternVL2.5-78B-MPO leads with 63.4% among models released by January 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is InternVL2.5-78B-MPO, at 63.4%. This view shows only models we track, each at its best.

The newest model with a score here came out in January 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 K12-PEM 69.4%
  2. 2 GPT-4o (snapshot unspecified) 63.9%
  3. 3 InternVL2.5-78B-MPO 63.4%
  4. 4 Qwen2.5-VL-72B 63.2%
  5. 5 InternVL2_5-78B 60.4%
  6. 6 Qwen2-VL-72B 54.9%
  7. 7 InternVL2.5-8B-MPO 53.2%
  8. 8 Qwen2.5-VL-7B 48.7%
  9. 9 InternVL2.5-8B 47%

Source: The K12-PEBench paper

#EntryScoreSettingDateSource
1 K12-PEM 69.4% Official leaderboard
2 GPT-4o (snapshot unspecified) 63.9% Official leaderboard
3 InternVL2.5-78B-MPO 63.4% Dec 2024 Official leaderboard
4 Qwen2.5-VL-72B 63.2% Jan 2025 Official leaderboard
5 InternVL2_5-78B 60.4% Dec 2024 Official leaderboard
6 Qwen2-VL-72B 54.9% Official leaderboard
7 InternVL2.5-8B-MPO 53.2% Dec 2024 Official leaderboard
8 Qwen2.5-VL-7B 48.7% Jan 2025 Official leaderboard
9 InternVL2.5-8B 47% Dec 2024 Official leaderboard
#ModelScoreSettingDateSource
1 InternVL2.5-78B-MPO 63.4% Dec 2024 Official leaderboard
2 Qwen2.5-VL-72B 63.2% Jan 2025 Official leaderboard
3 InternVL2_5-78B 60.4% Dec 2024 Official leaderboard
4 InternVL2.5-8B-MPO 53.2% Dec 2024 Official leaderboard
5 Qwen2.5-VL-7B 48.7% Jan 2025 Official leaderboard
6 InternVL2.5-8B 47% Dec 2024 Official leaderboard

Rankings by part

We read 10 parts from the benchmark’s paper, each ranking models on one part of the test, such as Overall, CS and ICE. Qwen2.5-VL-72B leads 5 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Overall

6 models we track have a score here.

  1. 1 InternVL2.5-78B-MPO 63.4%
  2. 2 Qwen2.5-VL-72B 63.2%
  3. 3 InternVL2_5-78B 60.4%
  4. 4 InternVL2.5-8B-MPO 53.2%
  5. 5 Qwen2.5-VL-7B 48.7%
  6. 6 InternVL2.5-8B 47%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Shanghai AI Lab
Benchmark paper Not tested since Jun 2025

Source: The K12-PEBench paper

Scores over time

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

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) InternVL2.5-78B-MPO InternVL2.5-8B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) InternVL2.5-78B-MPO InternVL2.5-8B

Source: Official K12-PEBench leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 637 in Law, finance and medicine
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
  • Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 22 months ago, in January 2025, for Qwen2.5-VL-72B. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (9 entries).
  • Pass Models scored 6 models we track, plus 3 other entries on its leaderboard.
  • Warning Newest score The newest came 22 months ago, in January 2025, for Qwen2.5-VL-72B.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 63.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 6 rows for models we track and 3 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check