Benchmark Almanac

AstroVisBench

Gemini 2.5 Pro (snapshot unspecified) 69.2% tops its leaderboard Claude Opus 4 49.8% leads the models we track
Not in model cards

Models write code to process astronomical observations and plot the results. The benchmark compares computed values with reference outputs and uses a model to grade plots.

The authors publish seven metrics for eight models. The 864 tasks come from 110 astronomy notebooks.

Released
May 2025
Kind
Lab and data work
Level
University
Scored
Graded by a model
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Code → Code
Interaction
Single turn
Environment
Code sandbox
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
Strongest model tested
Claude Opus 4 #109 on the capability index
Last model added
May 2025
Scores refreshed
November 20th, 2026
8
Entries on its own leaderboard, 7 models we track

Gemini 2.5 Pro (snapshot unspecified) tops the leaderboard with 69.2% Claude Opus 4 leads with 49.8% among models released by May 2025

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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 Opus 4, at 49.8%. This view shows only models we track, each at its best.

The newest model with a score here came out in May 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 Gemini 2.5 Pro (snapshot unspecified) 69.2%
  2. 2 Claude Opus 4 49.8%
  3. 3 Claude 3.7 Sonnet 49.1%
  4. 4 o3-mini 48.6%
  5. 5 GPT-4o Mar 2025 46.3%
  6. 6 Llama 4 Maverick 44.7%
  7. 7 Qwen2.5-72B 43.1%
  8. 8 QwQ-32B 35.9%

Source: Official AstroVisBench leaderboard

#EntryScoreSettingDateSource
1 Gemini 2.5 Pro (snapshot unspecified) 69.2% Official leaderboard
2 Claude Opus 4 49.8% May 2025 Official leaderboard
3 Claude 3.7 Sonnet 49.1% Feb 2025 Official leaderboard
4 o3-mini 48.6% Jan 2025 Official leaderboard
5 GPT-4o Mar 2025 46.3% Mar 2025 Official leaderboard
6 Llama 4 Maverick 44.7% Apr 2025 Official leaderboard
7 Qwen2.5-72B 43.1% Sep 2024 Official leaderboard
8 QwQ-32B 35.9% Mar 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Opus 4 49.8% May 2025 Official leaderboard
2 Claude 3.7 Sonnet 49.1% Feb 2025 Official leaderboard
3 o3-mini 48.6% Jan 2025 Official leaderboard
4 GPT-4o Mar 2025 46.3% Mar 2025 Official leaderboard
5 Llama 4 Maverick 44.7% Apr 2025 Official leaderboard
6 Qwen2.5-72B 43.1% Sep 2024 Official leaderboard
7 QwQ-32B 35.9% Mar 2025 Official leaderboard

Rankings by part

We read 7 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as NoErr(P) %, VIscore and NoErr(V) %. o3-mini leads 2 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

NoErr(P) %

7 models we track have a score here.

  1. 1 Claude Opus 4 49.8%
  2. 2 Claude 3.7 Sonnet 49.1%
  3. 3 o3-mini 48.6%
  4. 4 GPT-4o Mar 2025 46.3%
  5. 5 Llama 4 Maverick 44.7%
  6. 6 Qwen2.5-72B 43.1%
  7. 7 QwQ-32B 35.9%

2 models have held number one on the official leaderboard since its release in May 2025

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

Anthropic
Own leaderboard Not tested since May 2025

Source: Official AstroVisBench leaderboard

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
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Claude Opus 4 Claude 3.7 Sonnet Qwen2.5-72B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) Claude Opus 4 Qwen2.5-72B

Source: Official AstroVisBench 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

All 203 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 18 months ago, in May 2025, for Claude Opus 4. 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 official leaderboard (8 entries).
  • Pass Models scored 7 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 18 months ago, in May 2025, for Claude Opus 4.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 4 (#109).
  • Pass Headroom The best is 49.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on astrovisbench.github.io: 7 rows for models we track and 1 other entry 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