AstroMLab 1
Models answer 4,425 multiple choice questions drawn from astronomy review articles. The authors publish the questions and grade submissions with a private answer key.
The AstroMLab 4 paper reports 86.2% accuracy for AstroSage on the original questions and compares it with general purpose models.
- Made by
- AstroMLab
- Released
- July 2024
- Kind
- Science questions
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Subject
- Physical sciences
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- A host
- Strongest model tested
- Claude 3 Opus #173 on the capability index
- Last model added
- September 2024
- Scores refreshed
- November 20th, 2026
Claude-3.5-Sonnet (snapshot unspecified) tops the leaderboard with 85% Claude 3 Opus leads with 82.7% among models released by September 2024
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 Claude 3 Opus, at 82.7%. This view shows only models we track, each at its best.
The newest model with a score here came out in September 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
- 85.0 Claude-3.5-Sonnet (snapshot unspecified)
- 82.7 Claude 3 Opus
- 80.6
Llama 3-70B
- 80.4 GPT-4o (snapshot unspecified)
- 77.9 Claude 3 Haiku
- 77.7 Mixtral-8x22B-v0.1
- 77.7 Qwen-2-70B
- 77.6 Gemini-1.5-Pro (snapshot unspecified)
- 77.3
Yi-Large
- 76.7 Claude 3 Sonnet
- 76.6 Step-2
- 75.6 Phi-3-14B
- 75.6
Gemma 2 27B
- 75.3 Claude 2
- 75.2 Step-1
- 75.1 ERNIE-4.0
- 75.1 GLM-4-0520
- 74.5 GPT-4 (snapshot unspecified)
- 73.7 Mistral-8x7B-v0.1
- 73.6 Deepseek-v2
- 73.6 Gemini-1.5-Flash (snapshot unspecified)
- 73.2
Qwen1.5-32B
- 73.1
Yi-1.5-34B
- 72.9 GLM-4-Air
- 72.9
Llama 3-8B
- 72.7 ABAB-6.5
- 72.7
Qwen1.5-110B
- 72.5 GLM-4-AirX
- 72.3 Moonshot-v1
- 72.1 ERNIE-3.5
- 71.8
Qwen2.5-7B
- 71.7 Phi-3-4B
- 71.5
Gemma 2 9B
- 71.0
Gemini 1.0 Pro
- 70.7
Llama 2-70B
- 70.4 GPT-3.5 (snapshot unspecified)
- 70.3 Yi-Medium
- 70.1 Doubao-Pro
- 69.5 ABAB-5.5
- 68.4 Yi-1.5-9B
- 68.0
Qwen2-7B
- 67.7
Qwen1.5-14B
- 67.1 GLM-4-Flash
- 67.0 GLM-4-9B
- 65.6 Phi-2-3B
- 64.5 InternLM-2.5-7B
- 64.3 GLM-3-Turbo
- 63.9
Mistral 7B v0.3
- 63.7
Qwen1.5-7B
- 63.1 Deepseek-67B
- 62.1
Mistral 7B v0.2
- 61.0 Yi-1.5-6B
- 60.5 Doubao-Lite
- 57.4 Qwen-1-7B
- 56.1 Gemma-1-7B
- 50.4 ChatGLM3-6B
- 50.3
Llama 2-7B
- 48.1
Mistral 7B v0.1
- 44.1 Gemma-1-2B
- 1 Claude-3.5-Sonnet (snapshot unspecified) 85%
- 2 Claude 3 Opus 82.7%
- 3
Llama 3-70B 80.6%
- 4 GPT-4o (snapshot unspecified) 80.4%
- 5 Claude 3 Haiku 77.9%
- 6 Mixtral-8x22B-v0.1 77.7%
- 7 Qwen-2-70B 77.7%
- 8 Gemini-1.5-Pro (snapshot unspecified) 77.6%
- 9
Yi-Large 77.3%
- 10 Claude 3 Sonnet 76.7%
- 11 Step-2 76.6%
- 12 Phi-3-14B 75.6%
- 13
Gemma 2 27B 75.6%
- 14 Claude 2 75.3%
- 15 Step-1 75.2%
- 16 ERNIE-4.0 75.1%
- 17 GLM-4-0520 75.1%
- 18 GPT-4 (snapshot unspecified) 74.5%
- 19 Mistral-8x7B-v0.1 73.7%
- 20 Deepseek-v2 73.6%
- 21 Gemini-1.5-Flash (snapshot unspecified) 73.6%
- 22
Qwen1.5-32B 73.2%
- 23
Yi-1.5-34B 73.1%
- 24 GLM-4-Air 72.9%
- 25
Llama 3-8B 72.9%
- 26 ABAB-6.5 72.7%
- 27
Qwen1.5-110B 72.7%
- 28 GLM-4-AirX 72.5%
- 29 Moonshot-v1 72.3%
- 30 ERNIE-3.5 72.1%
- 31
Qwen2.5-7B 71.8%
- 32 Phi-3-4B 71.7%
- 33
Gemma 2 9B 71.5%
- 34
Gemini 1.0 Pro 71%
- 35
Llama 2-70B 70.7%
- 36 GPT-3.5 (snapshot unspecified) 70.4%
- 37 Yi-Medium 70.3%
- 38 Doubao-Pro 70.1%
- 39 ABAB-5.5 69.5%
- 40 Yi-1.5-9B 68.4%
- 41
Qwen2-7B 68%
- 42
Qwen1.5-14B 67.7%
- 43 GLM-4-Flash 67.1%
- 44 GLM-4-9B 67%
- 45 Phi-2-3B 65.6%
- 46 InternLM-2.5-7B 64.5%
- 47 GLM-3-Turbo 64.3%
- 48
Mistral 7B v0.3 63.9%
- 49
Qwen1.5-7B 63.7%
- 50 Deepseek-67B 63.1%
- 51
Mistral 7B v0.2 62.1%
- 52 Yi-1.5-6B 61%
- 53 Doubao-Lite 60.5%
- 54 Qwen-1-7B 57.4%
- 55 Gemma-1-7B 56.1%
- 56 ChatGLM3-6B 50.4%
- 57
Llama 2-7B 50.3%
- 58
Mistral 7B v0.1 48.1%
- 59 Gemma-1-2B 44.1%
Source: The AstroMLab 1 paper
- 82.7 Claude 3 Opus
- 80.6
Llama 3-70B
- 77.9 Claude 3 Haiku
- 77.3
Yi-Large
- 76.7 Claude 3 Sonnet
- 75.6
Gemma 2 27B
- 75.3 Claude 2
- 73.2
Qwen1.5-32B
- 73.1
Yi-1.5-34B
- 72.9
Llama 3-8B
- 72.7
Qwen1.5-110B
- 71.8
Qwen2.5-7B
- 71.5
Gemma 2 9B
- 71.0
Gemini 1.0 Pro
- 70.7
Llama 2-70B
- 68.0
Qwen2-7B
- 67.7
Qwen1.5-14B
- 67.0 GLM-4-9B
- 63.9
Mistral 7B v0.3
- 63.7
Qwen1.5-7B
- 62.1
Mistral 7B v0.2
- 50.4 ChatGLM3-6B
- 50.3
Llama 2-7B
- 48.1
Mistral 7B v0.1
- 1 Claude 3 Opus 82.7%
- 2
Llama 3-70B 80.6%
- 3 Claude 3 Haiku 77.9%
- 4
Yi-Large 77.3%
- 5 Claude 3 Sonnet 76.7%
- 6
Gemma 2 27B 75.6%
- 7 Claude 2 75.3%
- 8
Qwen1.5-32B 73.2%
- 9
Yi-1.5-34B 73.1%
- 10
Llama 3-8B 72.9%
- 11
Qwen1.5-110B 72.7%
- 12
Qwen2.5-7B 71.8%
- 13
Gemma 2 9B 71.5%
- 14
Gemini 1.0 Pro 71%
- 15
Llama 2-70B 70.7%
- 16
Qwen2-7B 68%
- 17
Qwen1.5-14B 67.7%
- 18 GLM-4-9B 67%
- 19
Mistral 7B v0.3 63.9%
- 20
Qwen1.5-7B 63.7%
- 21
Mistral 7B v0.2 62.1%
- 22 ChatGLM3-6B 50.4%
- 23
Llama 2-7B 50.3%
- 24
Mistral 7B v0.1 48.1%
Source: The AstroMLab 1 paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude-3.5-Sonnet (snapshot unspecified) | 85% | Official leaderboard | ||
| 2 | Claude 3 Opus | 82.7% | Feb 2024 | Official leaderboard | |
| 3 | | 80.6% | Apr 2024 | Official leaderboard | |
| 4 | GPT-4o (snapshot unspecified) | 80.4% | Official leaderboard | ||
| 5 | Claude 3 Haiku | 77.9% | Mar 2024 | Official leaderboard | |
| 6 | Mixtral-8x22B-v0.1 | 77.7% | Official leaderboard | ||
| 7 | Qwen-2-70B | 77.7% | Official leaderboard | ||
| 8 | Gemini-1.5-Pro (snapshot unspecified) | 77.6% | Official leaderboard | ||
| 9 | | 77.3% | May 2024 | Official leaderboard | |
| 10 | Claude 3 Sonnet | 76.7% | Feb 2024 | Official leaderboard | |
| 11 | Step-2 | 76.6% | Official leaderboard | ||
| 12 | Phi-3-14B | 75.6% | Official leaderboard | ||
| 13 | | 75.6% | Jun 2024 | Official leaderboard | |
| 14 | Claude 2 | 75.3% | Jul 2023 | Official leaderboard | |
| 15 | Step-1 | 75.2% | Official leaderboard | ||
| 16 | ERNIE-4.0 | 75.1% | Official leaderboard | ||
| 17 | GLM-4-0520 | 75.1% | Official leaderboard | ||
| 18 | GPT-4 (snapshot unspecified) | 74.5% | Official leaderboard | ||
| 19 | Mistral-8x7B-v0.1 | 73.7% | Official leaderboard | ||
| 20 | Deepseek-v2 | 73.6% | Official leaderboard | ||
| 21 | Gemini-1.5-Flash (snapshot unspecified) | 73.6% | Official leaderboard | ||
| 22 | | 73.2% | Feb 2024 | Official leaderboard | |
| 23 | | 73.1% | May 2024 | Official leaderboard | |
| 24 | GLM-4-Air | 72.9% | Official leaderboard | ||
| 25 | | 72.9% | Apr 2024 | Official leaderboard | |
| 26 | ABAB-6.5 | 72.7% | Official leaderboard | ||
| 27 | | 72.7% | Apr 2024 | Official leaderboard | |
| 28 | GLM-4-AirX | 72.5% | Official leaderboard | ||
| 29 | Moonshot-v1 | 72.3% | Official leaderboard | ||
| 30 | ERNIE-3.5 | 72.1% | Official leaderboard | ||
| 31 | | 71.8% | Sep 2024 | Official leaderboard | |
| 32 | Phi-3-4B | 71.7% | Official leaderboard | ||
| 33 | | 71.5% | Jun 2024 | Official leaderboard | |
| 34 | | 71% | Dec 2023 | Official leaderboard | |
| 35 | | 70.7% | Jul 2023 | Official leaderboard | |
| 36 | GPT-3.5 (snapshot unspecified) | 70.4% | Official leaderboard | ||
| 37 | Yi-Medium | 70.3% | Official leaderboard | ||
| 38 | Doubao-Pro | 70.1% | Official leaderboard | ||
| 39 | ABAB-5.5 | 69.5% | Official leaderboard | ||
| 40 | Yi-1.5-9B | 68.4% | Official leaderboard | ||
| 41 | | 68% | Jun 2024 | Official leaderboard | |
| 42 | | 67.7% | Feb 2024 | Official leaderboard | |
| 43 | GLM-4-Flash | 67.1% | Official leaderboard | ||
| 44 | GLM-4-9B | 67% | Jun 2024 | Official leaderboard | |
| 45 | Phi-2-3B | 65.6% | Official leaderboard | ||
| 46 | InternLM-2.5-7B | 64.5% | Official leaderboard | ||
| 47 | GLM-3-Turbo | 64.3% | Official leaderboard | ||
| 48 | | 63.9% | Sep 2023 | Official leaderboard | |
| 49 | | 63.7% | Feb 2024 | Official leaderboard | |
| 50 | Deepseek-67B | 63.1% | Official leaderboard | ||
| 51 | | 62.1% | Dec 2023 | Official leaderboard | |
| 52 | Yi-1.5-6B | 61% | Official leaderboard | ||
| 53 | Doubao-Lite | 60.5% | Official leaderboard | ||
| 54 | Qwen-1-7B | 57.4% | Official leaderboard | ||
| 55 | Gemma-1-7B | 56.1% | Official leaderboard | ||
| 56 | ChatGLM3-6B | 50.4% | Oct 2023 | Official leaderboard | |
| 57 | | 50.3% | Jul 2023 | Official leaderboard | |
| 58 | | 48.1% | Sep 2023 | Official leaderboard | |
| 59 | Gemma-1-2B | 44.1% | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude 3 Opus | 82.7% | Feb 2024 | Official leaderboard | |
| 2 | | 80.6% | Apr 2024 | Official leaderboard | |
| 3 | Claude 3 Haiku | 77.9% | Mar 2024 | Official leaderboard | |
| 4 | | 77.3% | May 2024 | Official leaderboard | |
| 5 | Claude 3 Sonnet | 76.7% | Feb 2024 | Official leaderboard | |
| 6 | | 75.6% | Jun 2024 | Official leaderboard | |
| 7 | Claude 2 | 75.3% | Jul 2023 | Official leaderboard | |
| 8 | | 73.2% | Feb 2024 | Official leaderboard | |
| 9 | | 73.1% | May 2024 | Official leaderboard | |
| 10 | | 72.9% | Apr 2024 | Official leaderboard | |
| 11 | | 72.7% | Apr 2024 | Official leaderboard | |
| 12 | | 71.8% | Sep 2024 | Official leaderboard | |
| 13 | | 71.5% | Jun 2024 | Official leaderboard | |
| 14 | | 71% | Dec 2023 | Official leaderboard | |
| 15 | | 70.7% | Jul 2023 | Official leaderboard | |
| 16 | | 68% | Jun 2024 | Official leaderboard | |
| 17 | | 67.7% | Feb 2024 | Official leaderboard | |
| 18 | GLM-4-9B | 67% | Jun 2024 | Official leaderboard | |
| 19 | | 63.9% | Sep 2023 | Official leaderboard | |
| 20 | | 63.7% | Feb 2024 | Official leaderboard | |
| 21 | | 62.1% | Dec 2023 | Official leaderboard | |
| 22 | ChatGLM3-6B | 50.4% | Oct 2023 | Official leaderboard | |
| 23 | | 50.3% | Jul 2023 | Official leaderboard | |
| 24 | | 48.1% | Sep 2023 | Official leaderboard |
Rankings by part
We read 11 parts from the benchmark’s paper in 2 groups: subfield and tested ability. Each ranks models on one part of the test. Claude 3 Opus leads 10 of them, and Llama 3-70B leads the other one. Arrows show how far a model moved from its overall rank among the models we track.
Solar & Stellar
24 models we track have a score here.
- 1 Claude 3 Opus 81.5%
- 2
Llama 3-70B 78.7%
- 3 Claude 3 Haiku 77.6%
- 4 Claude 3 Sonnet ↑1 76.1%
- 5
Yi-Large ↓1 75.6%
- 6 Claude 2 ↑1 74.1%
- 7
Gemma 2 27B ↓1 73.3%
- 8
Llama 3-8B ↑2 71.2%
- 9
Qwen1.5-110B ↑2 71.2%
- 10
Gemma 2 9B ↑3 71%
Earth & Planetary
24 models we track have a score here.
- 1 Claude 3 Opus 83.9%
- 2
Llama 3-70B 79.1%
- 3
Gemma 2 27B ↑3 77.8%
- 4
Yi-Large 75.9%
- 5 Claude 3 Haiku ↓2 74.3%
- 6 Claude 3 Sonnet ↓1 73.7%
- 7 Claude 2 72.4%
- 8
Qwen1.5-110B ↑3 71.5%
- 9
Gemini 1.0 Pro ↑5 71.3%
- 10
Llama 3-8B 70.7%
Galactic Astrophysics
24 models we track have a score here.
- 1 Claude 3 Opus 83.6%
- 2
Llama 3-70B 82.1%
- 3 Claude 3 Haiku 79.3%
- 4
Yi-Large 78.9%
- 5 Claude 2 ↑2 77.8%
- 6 Claude 3 Sonnet ↓1 77.6%
- 7
Gemma 2 27B ↓1 77.6%
- 8
Yi-1.5-34B ↑1 75.8%
- 9
Qwen1.5-32B ↓1 75.5%
- 10
Qwen2.5-7B ↑2 73.5%
Cosmology & Nongalactic
24 models we track have a score here.
- 1 Claude 3 Opus 84.1%
- 2
Llama 3-70B 81.5%
- 3 Claude 3 Haiku 78.2%
- 4 Claude 3 Sonnet ↑1 76.4%
- 5
Yi-Large ↓1 76%
- 6
Gemma 2 27B 75.6%
- 7
Qwen2.5-7B ↑5 74.6%
- 8
Llama 3-8B ↑2 74.2%
- 9
Yi-1.5-34B 74.2%
- 10
Llama 2-70B ↑5 73.4%
High Energy Astrophysics
24 models we track have a score here.
- 1
Llama 3-70B ↑1 83.4%
- 2 Claude 3 Opus ↓1 83.2%
- 3 Claude 3 Haiku 80.9%
- 4 Claude 3 Sonnet ↑1 79.9%
- 5
Yi-Large ↓1 79.6%
- 6
Qwen1.5-32B ↑2 78.8%
- 7
Yi-1.5-34B ↑2 77.8%
- 8 Claude 2 ↓1 76.9%
- 9
Gemma 2 27B ↓3 76.8%
- 10
Llama 3-8B 76.5%
Instrumentation & Methods
24 models we track have a score here.
- 1 Claude 3 Opus 81.9%
- 2
Llama 3-70B 80.2%
- 3 Claude 2 ↑4 76.6%
- 4 Claude 3 Sonnet ↑1 76.3%
- 5
Yi-Large ↓1 76.3%
- 6 Claude 3 Haiku ↓3 75.4%
- 7
Llama 3-8B ↑3 75.1%
- 8
Llama 2-70B ↑7 74%
- 9
Gemma 2 27B ↓3 73.4%
- 10
Qwen1.5-32B ↓2 73.1%
Understanding Fundamental Concepts
24 models we track have a score here.
- 1 Claude 3 Opus 84.8%
- 2
Llama 3-70B 84.1%
- 3 Claude 3 Sonnet ↑2 81.7%
- 4 Claude 3 Haiku ↓1 81.1%
- 5
Yi-1.5-34B ↑4 80.1%
- 6
Yi-Large ↓2 79.5%
- 7 Claude 2 78.5%
- 8
Qwen1.5-110B ↑3 78.4%
- 9
Qwen2.5-7B ↑3 78.2%
- 10
Gemma 2 27B ↓4 78.1%
Technical & Observational Techniques
24 models we track have a score here.
- 1 Claude 3 Opus 82.1%
- 2
Llama 3-70B 82.1%
- 3
Yi-Large ↑1 80%
- 4 Claude 3 Sonnet ↑1 78.5%
- 5 Claude 3 Haiku ↓2 77.1%
- 6 Claude 2 ↑1 77%
- 7
Yi-1.5-34B ↑2 76.6%
- 8
Qwen1.5-110B ↑3 76.2%
- 9
Gemma 2 27B ↓3 75.6%
- 10
Llama 3-8B 75%
Analytical & Reasoning Skills
24 models we track have a score here.
- 1 Claude 3 Opus 76.4%
- 2
Qwen1.5-32B ↑6 73.6%
- 3 Claude 3 Sonnet ↑2 73.6%
- 4
Llama 3-70B ↓2 73.6%
- 5
Qwen1.5-110B ↑6 72.2%
- 6
Gemma 2 27B 72.2%
- 7 Claude 2 70.8%
- 8
Yi-Large ↓4 70.8%
- 9
Qwen2-7B ↑7 70.4%
- 10
Qwen1.5-14B ↑7 69.6%
Historical & Theoretical Knowledge
24 models we track have a score here.
- 1 Claude 3 Opus 82.5%
- 2
Llama 3-70B 78.9%
- 3
Gemma 2 27B ↑3 77.7%
- 4 Claude 3 Haiku ↓1 75.9%
- 5
Llama 3-8B ↑5 75.3%
- 6 Claude 3 Sonnet ↓1 73.5%
- 7
Yi-Large ↓3 72.9%
- 8
Qwen2.5-7B ↑4 72.9%
- 9 Claude 2 ↓2 71.1%
- 10
Gemini 1.0 Pro ↑4 71.1%
Advanced Topics
24 models we track have a score here.
- 1 Claude 3 Opus 87%
- 2
Llama 3-70B 84%
- 3
Yi-Large ↑1 84%
- 4 Claude 3 Haiku ↓1 81.5%
- 5
Qwen1.5-32B ↑3 81.3%
- 6
Gemma 2 27B 80.9%
- 7
Qwen1.5-110B ↑4 80.3%
- 8 Claude 3 Sonnet ↓3 79.6%
- 9 Claude 2 ↓2 77.2%
- 10
Yi-1.5-34B ↓1 75.9%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The AstroMLab 1 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 AstroMLab 1 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 Science and reasoning
- GPQA Diamond In use 198 graduate-level 'Google-proof' multiple-choice questions in biology, chemistry and physics.
Gemini 4 Carbon 97%
- Terminal-Bench-Science In use 70 scientific research-workflow tasks across five science areas, run in Harbor containers. Hidden tests give the resolution rate. GPT-6 Astra 68.1%
- BioMysteryBench In use Bioinformatics problems on anonymized real datasets that require real analysis with tools. Rubric-graded, with human-solvable and human-difficult splits. Claude Fable 5.2 83.7%
- FrontierScience In use Models solve physics, chemistry, and biology problems in an Olympiad track graded by short answer and a Research track graded by 10-point rubrics.
Intern-S2-397B 87%
- HiPhO In use Models solve theory problems from 2024 and 2025 physics olympiad exams, some with diagrams. Official marking schemes grade answers and steps, mapped to medal levels.
Gemini 3 Pro 78.3%
- LABBench2 In use 11 biology research sub-tasks (literature, figures, tables, patents, trials and more) with tool access. Macro-averaged accuracy. GPT-5.2 60.9%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in September 2024, for Qwen2.5-7B. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (59 entries).
- Pass Models scored 24 models we track, plus 35 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in September 2024, for Qwen2.5-7B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude 3 Opus (#173).
- Pass Headroom The best is 82.7%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 24 rows for models we track and 35 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.