RACE
InternVL3.5-241B-A28B 94.2% leads since August 2025 Models answer reading comprehension questions from Chinese middle and high school English exams. Reported here without a subset split.
Use race-m or race-h when a report gives the middle or high school subset.
- Made by
- Carnegie Mellon University
- Released
- April 2017
- Kind
- Reading comprehension
- Level
- School
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Strongest model tested
-
Llama 3.1-405B #162 on the capability index - Last in a system card
- October 2026
- Last model added
- October 2026
In the authors’ words
We present RACE, a new dataset for benchmark evaluation of methods in the reading comprehension task. Collected from the English exams for middle and high school Chinese students in the age range between 12 to 18, RACE consists of near 28,000 passages and near 100,000 questions generated by human experts (English instructors), and covers a variety of topics which are carefully designed for evaluating the students' ability in understanding and reasoning. In particular, the proportion of questions that requires reasoning is much larger in RACE than that in other benchmark datasets for reading comprehension, and there is a significant gap between the performance of the state-of-the-art models (43%) and the ceiling human performance (95%). We hope this new dataset can serve as a valuable resource for research and evaluation in machine comprehension. The dataset is freely available at http://www.cs.cmu.edu/~glai1/data/race/ and the code is available at https://github.com/qizhex/RACE_AR_baselines.
InternVL3.5-241B-A28B leads with 94.2%
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.
- 1 InternVL3.5-241B-A28B 94.2%
- 2
Nemotron 3 Ultra 550B-A55B Base 92.2% - 3 Nemotron 3.5 Super 120B-A12B Base 91.4%
- 4 Nemotron 3 Nano 30B-A3B Base 88%
- 5 Nemotron 3.5 Lightning 30B-A3B Base 87.8%
- 6 Llama 3.1-70B 59%
- 7 Llama 3.1-405B 58.1%
- 8 Llama 3.1-8B 54.3%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | InternVL3.5-241B-A28B | 94.2% | Aug 2025 | Lab model card | |
| 2 | Nemotron 3 Ultra 550B-A55B Base | 92.2% | Jun 2026 | Lab model card | |
| 3 | Nemotron 3.5 Super 120B-A12B Base | 91.4% | Oct 2026 | Lab model card | |
| 4 | Nemotron 3 Nano 30B-A3B Base | 88% | Dec 2025 | Lab model card | |
| 5 | Nemotron 3.5 Lightning 30B-A3B Base | 87.8% | Oct 2026 | Lab model card | |
| 6 | Llama 3.1-70B | 59% | Nov 2024 | Lab model card | |
| 7 | Llama 3.1-405B | 58.1% | Nov 2024 | Lab model card | |
| 8 | Llama 3.1-8B | 54.3% | Nov 2024 | Lab model card |
2 models have held number one in labs’ own numbers since 2023
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Labs’ model cards
Shanghai AI Lab’s own number for InternVL3.5-241B-A28B passed 85% in August 2025
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Labs’ model cards
Labs report it only in system cards
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.
8 system cards report it, from 3 labs
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 |
|---|---|---|---|---|
| Oct 15, 2026 | Nemotron 3.5 Super 120B-A12B Base technical report | 91.4 | RACE, 0-shot acc, Table 6 | Capabilities |
| Oct 15, 2026 | Nemotron 3.5 Lightning 30B-A3B Base technical report | 87.8 | RACE, 0-shot acc, Table 6 | Capabilities |
| Jun 4, 2026 | Nemotron 3 Ultra 550B-A55B Base model card | 92.2 | RACE, 0-shot acc | Capabilities |
| Dec 23, 2025 | Nemotron 3 Nano 30B-A3B Base technical report | 88 | RACE, 0-shot accuracy, base model Table 2 | Capabilities |
| Aug 25, 2025 | InternVL3.5-241B-A28B technical report | 94.2 | RACE-High, OpenCompass (Table 14) | Capabilities |
| Nov 23, 2024 | Llama 3.1-8B technical report | 54.3 | RACE, pretrained (Table 9) | Capabilities |
| Nov 23, 2024 | Llama 3.1-70B technical report | 59 | RACE, pretrained (Table 9) | Capabilities |
| Nov 23, 2024 | Llama 3.1-405B technical report | 58.1 | RACE, pretrained (Table 9) | Capabilities |
More in Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects. Intern-S2-397B 89.8%
- SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy.
Qwen 3.6 Max (Preview) 73.9% - WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1.
Tencent Hy4 preview 83.9% - BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly.
Gemma 4 31B IT 74.4% - KMMLU-Pro In use Models answer multiple-choice questions taken from Korean National Professional Licensure exams. It tests professional knowledge used in Korean industry.
Solar Pro 4 79.2% - HLE-Verified In use Verified and revised version of Humanity's Last Exam (1,811 items) with uncertain items removed.
Seed2.0 Pro 73.6%