AutoBencher
Models answer generated benchmark questions. The framework searches for topics and produces tests of knowledge, multilingual reasoning, and math.
- Made by
- Xiang Lisa Li and colleagues
- Released
- July 2024
- Kind
- Language reasoning
- Level
- University
- Scored
- Checked automatically
- Answers
- Short answer
- Tasks
- All public
- Measures
- Capability
- Subject
- Mathematics
- Language
- Many languages
- Interaction
- Single turn
- Items from
- Generated
- Refresh
- One fixed set
- Run by
- Anyone
- In views
- Many languages
- Strongest model tested
- GPT-4 Turbo Apr 2024 #172 on the capability index
- Last model added
- April 2024
- Scores refreshed
- November 20th, 2026
2 models share the top score of 55% among models released by April 2024 2 models share the top score of 55% among models released by April 2024
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The headline is the plain mean of the three generated knowledge tests in Table 11. The paper gives no accuracy across all AutoBencher evaluations. 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. This view shows only models we track, each at its best.
The newest model with a score here came out in April 2024, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 55.0 Claude 3 Opus
- 55.0 GPT-4 Turbo
- 53.0 Claude 3 Sonnet
- 51.0 GPT-3.5 Turbo
- 49.3 openchat-3.5-0106
- 48.7 Claude 2
- 47.0 OpenAGI-7B-v0.1
- 46.0 Mixtral-8x7B-Instruct-v0.1
- 44.3 vicuna-7b-v1.5
- 43.3 alpaca-7b
- 43.0
Mistral-7B-Instruct-v0.1
- 41.7 Llama-2-7b-chat-hf
- 37.7
Zephyr 7B Beta
- 35.0 WizardMath-7B-V1.0
- 34.0
Gemini 1.0 Pro
- 29.3 Xwin-Math-7B-V1.0
- 20.3 GPT-Neo-2.7B
- 1 Claude 3 Opus 55%
- 2 GPT-4 Turbo Apr 2024 55%
- 3 Claude 3 Sonnet 53%
- 4 GPT-3.5 Turbo Jun 2023 51%
- 5 openchat-3.5-0106 49.3%
- 6 Claude 2 48.7%
- 7 OpenAGI-7B-v0.1 47%
- 8 Mixtral-8x7B-Instruct-v0.1 46%
- 9 vicuna-7b-v1.5 44.3%
- 10 alpaca-7b 43.3%
- 11
Mistral-7B-Instruct-v0.1 43%
- 12 Llama-2-7b-chat-hf 41.7%
- 13
Zephyr 7B Beta 37.7%
- 14 WizardMath-7B-V1.0 35%
- 15
Gemini 1.0 Pro 34%
- 16 Xwin-Math-7B-V1.0 29.3%
- 17 GPT-Neo-2.7B 20.3%
Source: The AutoBencher paper
- 1 Claude 3 Opus 55%
- 2 GPT-4 Turbo Apr 2024 55%
- 3 Claude 3 Sonnet 53%
- 4 GPT-3.5 Turbo Jun 2023 51%
- 5 Claude 2 48.7%
- 6
Mistral-7B-Instruct-v0.1 43%
- 7
Zephyr 7B Beta 37.7%
- 8
Gemini 1.0 Pro 34%
- 9 GPT-Neo-2.7B 20.3%
Source: The AutoBencher paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3 Opus | 55% | Feb 2024 | Official leaderboard | |
| 2 | GPT-4 Turbo Apr 2024 | 55% | Apr 2024 | Official leaderboard | |
| 3 | Claude 3 Sonnet | 53% | Feb 2024 | Official leaderboard | |
| 4 | GPT-3.5 Turbo Jun 2023 | 51% | Jun 2023 | Official leaderboard | |
| 5 | openchat-3.5-0106 | 49.3% | Official leaderboard | ||
| 6 | Claude 2 | 48.7% | Jul 2023 | Official leaderboard | |
| 7 | OpenAGI-7B-v0.1 | 47% | Official leaderboard | ||
| 8 | Mixtral-8x7B-Instruct-v0.1 | 46% | Official leaderboard | ||
| 9 | vicuna-7b-v1.5 | 44.3% | Official leaderboard | ||
| 10 | alpaca-7b | 43.3% | Official leaderboard | ||
| 11 | | 43% | Sep 2023 | Official leaderboard | |
| 12 | Llama-2-7b-chat-hf | 41.7% | Official leaderboard | ||
| 13 | | 37.7% | Oct 2023 | Official leaderboard | |
| 14 | WizardMath-7B-V1.0 | 35% | Official leaderboard | ||
| 15 | | 34% | Dec 2023 | Official leaderboard | |
| 16 | Xwin-Math-7B-V1.0 | 29.3% | Official leaderboard | ||
| 17 | GPT-Neo-2.7B | 20.3% | Mar 2021 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3 Opus | 55% | Feb 2024 | Official leaderboard | |
| 2 | GPT-4 Turbo Apr 2024 | 55% | Apr 2024 | Official leaderboard | |
| 3 | Claude 3 Sonnet | 53% | Feb 2024 | Official leaderboard | |
| 4 | GPT-3.5 Turbo Jun 2023 | 51% | Jun 2023 | Official leaderboard | |
| 5 | Claude 2 | 48.7% | Jul 2023 | Official leaderboard | |
| 6 | | 43% | Sep 2023 | Official leaderboard | |
| 7 | | 37.7% | Oct 2023 | Official leaderboard | |
| 8 | | 34% | Dec 2023 | Official leaderboard | |
| 9 | GPT-Neo-2.7B | 20.3% | Mar 2021 | Official leaderboard |
Rankings by part
We read 3 parts from the benchmark’s paper: History, Economy and Science. Each ranks models on one part of the test. GPT-4 Turbo leads 1 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.
History
9 models we track have a score here.
- 1 GPT-4 Turbo Apr 2024 ↑1 53%
- 2 GPT-3.5 Turbo Jun 2023 ↑2 51%
- 3 Claude 3 Opus ↓2 51%
- 4 Claude 2 ↑1 42%
- 5 Claude 3 Sonnet ↓2 42%
- 6
Mistral-7B-Instruct-v0.1 37%
- 7
Zephyr 7B Beta 35%
- 8
Gemini 1.0 Pro 28%
- 9 GPT-Neo-2.7B 26%
Economy
9 models we track have a score here.
- 1 Claude 3 Sonnet ↑2 67%
- 2 Claude 3 Opus ↓1 64%
- 3 Claude 2 ↑2 62%
- 4 GPT-4 Turbo Apr 2024 ↓2 62%
- 5 GPT-3.5 Turbo Jun 2023 ↓1 60%
- 6
Mistral-7B-Instruct-v0.1 57%
- 7
Zephyr 7B Beta 48%
- 8
Gemini 1.0 Pro 48%
- 9 GPT-Neo-2.7B 26%
Science
9 models we track have a score here.
- 1 Claude 3 Opus 50%
- 2 Claude 3 Sonnet ↑1 50%
- 3 GPT-4 Turbo Apr 2024 ↓1 50%
- 4 GPT-3.5 Turbo Jun 2023 42%
- 5 Claude 2 42%
- 6
Mistral-7B-Instruct-v0.1 35%
- 7
Zephyr 7B Beta 30%
- 8
Gemini 1.0 Pro 26%
- 9 GPT-Neo-2.7B 9.00%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The AutoBencher paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official AutoBencher 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 Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects.
Intern-S2-397B 89.8%
- MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy.
Seed1.8 92.3%
- 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%
- HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in April 2024, for GPT-4 Turbo. 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 (17 entries).
- Pass Models scored 9 models we track, plus 8 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in April 2024, for GPT-4 Turbo.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-4 Turbo (#172).
- Pass Headroom The best is 55%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 9 rows for models we track and 8 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.