Benchmark Almanac

LLM Physics Plotting Benchmark

GPT-5.2 Codex 90.6% leads since December 2025
Not in model cards

Models write Python code to produce requested plots from ATLAS collision data. Each trial passes when the code runs and meets the configured checks.

The default workflow reports 3,138 trials across 20 model configurations and 18 questions. The report includes costs and separates plotting workflows.

Made by
Gordon Watts
Released
September 2025
Kind
Lab and data work
Level
University
Scored
Checked automatically
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Tables → Code
Interaction
Single turn
Environment
Code sandbox
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
Strongest model tested
GPT-5.6 Sol #14 on the capability index
Last model added
July 2026
Scores refreshed
November 20th, 2026
19
Entries on its own leaderboard, 17 models we track

GPT-5.2 Codex leads with 90.6% among models released by July 2026 GPT-5.2 Codex leads with 90.6% among models released by July 2026

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. Every run is in the next chart. 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 July 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.2 Codex ServiceX + Awkward Array + Hist · Combined current + historical contexts 90.6%
  2. 2 GPT-5.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6%
  3. 3 GPT-5.6 Luna ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6%
  4. 4 GPT-5.6 Sol ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6%
  5. 5 GPT-5 mini ServiceX + Awkward Array + Hist · Combined current + historical contexts 85%
  6. 6 kimi-k27-nrp ServiceX + Awkward Array + Hist · Combined current + historical contexts 84.6%
  7. 7 GLM-5 ServiceX + ROOT RDataFrame · Combined recorded contexts 83%
  8. 8 Gemini 2.5 Pro Jun 2025, ServiceX + Awkward Array + Hist · Combined current + historical contexts 82.2%
  9. 9 Kimi K2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 80.6%
  10. 10 Claude Sonnet 4.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 74.4%
  11. 11 Claude Opus 4.6 ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.9%
  12. 12 Gemini 3 Flash ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.3%
  13. 13 MiniMax-M2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 65%
  14. 14 Qwen3.5-nrp ServiceX + Awkward Array + Hist · Combined current + historical contexts 61.1%
  15. 15 gpt-oss-120b ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6%
  16. 16 grok-code-fast-1 ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6%
  17. 17 DeepSeek-V3.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 53.9%
  18. 18 gpt-oss-20b ServiceX + Awkward Array + Hist · Combined current + historical contexts 45%
  19. 19 Llama 3.3 70B ServiceX + Awkward Array + Hist · Combined current + historical contexts 25%

Source: Official LLM Physics Plotting Benchmark leaderboard

  1. 1 GPT-5.2 Codex ServiceX + Awkward Array + Hist · Combined current + historical contexts 90.6%
  2. 2 GPT-5.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6%
  3. 3 GPT-5.6 Luna ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6%
  4. 4 GPT-5.6 Sol ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6%
  5. 5 GPT-5 mini ServiceX + Awkward Array + Hist · Combined current + historical contexts 85%
  6. 6 GLM-5 ServiceX + ROOT RDataFrame · Combined recorded contexts 83%
  7. 7 Gemini 2.5 Pro Jun 2025, ServiceX + Awkward Array + Hist · Combined current + historical contexts 82.2%
  8. 8 Kimi K2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 80.6%
  9. 9 Claude Sonnet 4.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 74.4%
  10. 10 Claude Opus 4.6 ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.9%
  11. 11 Gemini 3 Flash ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.3%
  12. 12 MiniMax-M2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 65%
  13. 13 gpt-oss-120b ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6%
  14. 14 grok-code-fast-1 ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6%
  15. 15 DeepSeek-V3.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 53.9%
  16. 16 gpt-oss-20b ServiceX + Awkward Array + Hist · Combined current + historical contexts 45%
  17. 17 Llama 3.3 70B ServiceX + Awkward Array + Hist · Combined current + historical contexts 25%

Source: Official LLM Physics Plotting Benchmark leaderboard

#EntryScoreSettingDateSource
1 GPT-5.2 Codex 90.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
2 GPT-5.2 85.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
3 GPT-5.6 Luna 85.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jul 2026 Official leaderboard
4 GPT-5.6 Sol 85.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jul 2026 Official leaderboard
5 GPT-5 mini 85% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
6 kimi-k27-nrp 84.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Official leaderboard
7 GLM-5 83% ServiceX + ROOT RDataFrame · Combined recorded contexts Feb 2026 Official leaderboard
8 Gemini 2.5 Pro Jun 2025 82.2% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jun 2025 Official leaderboard
9 Kimi K2.5 80.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jan 2026 Official leaderboard
10 Claude Sonnet 4.5 74.4% ServiceX + Awkward Array + Hist · Combined current + historical contexts Sep 2025 Official leaderboard
11 Claude Opus 4.6 73.9% ServiceX + Awkward Array + Hist · Combined current + historical contexts Feb 2026 Official leaderboard
12 Gemini 3 Flash 73.3% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
13 MiniMax-M2.5 65% ServiceX + Awkward Array + Hist · Combined current + historical contexts Feb 2026 Official leaderboard
14 Qwen3.5-nrp 61.1% ServiceX + Awkward Array + Hist · Combined current + historical contexts Official leaderboard
15 gpt-oss-120b 60.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
16 grok-code-fast-1 60.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
17 DeepSeek-V3.2 53.9% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
18 gpt-oss-20b 45% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
19 Llama 3.3 70B 25% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2024 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5.2 Codex 90.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
2 GPT-5.2 85.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
3 GPT-5.6 Luna 85.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jul 2026 Official leaderboard
4 GPT-5.6 Sol 85.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jul 2026 Official leaderboard
5 GPT-5 mini 85% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
6 GLM-5 83% ServiceX + ROOT RDataFrame · Combined recorded contexts Feb 2026 Official leaderboard
7 Gemini 2.5 Pro Jun 2025 82.2% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jun 2025 Official leaderboard
8 Kimi K2.5 80.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Jan 2026 Official leaderboard
9 Claude Sonnet 4.5 74.4% ServiceX + Awkward Array + Hist · Combined current + historical contexts Sep 2025 Official leaderboard
10 Claude Opus 4.6 73.9% ServiceX + Awkward Array + Hist · Combined current + historical contexts Feb 2026 Official leaderboard
11 Gemini 3 Flash 73.3% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
12 MiniMax-M2.5 65% ServiceX + Awkward Array + Hist · Combined current + historical contexts Feb 2026 Official leaderboard
13 gpt-oss-120b 60.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
14 grok-code-fast-1 60.6% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
15 DeepSeek-V3.2 53.9% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2025 Official leaderboard
16 gpt-oss-20b 45% ServiceX + Awkward Array + Hist · Combined current + historical contexts Aug 2025 Official leaderboard
17 Llama 3.3 70B 25% ServiceX + Awkward Array + Hist · Combined current + historical contexts Dec 2024 Official leaderboard

Rankings by part

We read 18 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Q1, Q2 and Q3. Gemini 2.5 Pro leads 8 of them, and 6 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Q1

17 models we track have a score here.

  1. 1 Gemini 2.5 Pro Jun 2025 ↑6 100%
  2. 2 gpt-oss-120b ↑11 100%
  3. 3 gpt-oss-20b ↑13 100%
  4. 4 GPT-5 mini ↑1 100%
  5. 5 grok-code-fast-1 ↑9 100%
  6. 6 Claude Sonnet 4.5 ↑3 100%
  7. 7 DeepSeek-V3.2 ↑8 100%
  8. 8 GPT-5.2 ↓6 100%
  9. 9 Gemini 3 Flash ↑2 100%
  10. 10 GPT-5.2 Codex ↓9 100%

The best score, 90.6%, costs $0.10 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-5.6 Luna at ServiceX + Awkward Array + Hist · Combined current + historical contexts comes within 5 points of the best for $0.03 a task, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.01 $0.02 $0.05 $0.10 $0.20 $0.50 $1.00 Cost per task, log scale GPT-5.2 Codex, ServiceX + Awkward Array + Hist · Combined current + historical contexts GPT-5.6 Luna, ServiceX + Awkward Array + Hist · Combined current + historical contexts GPT-5 mini, ServiceX + ROOT RDataFrame · Combined recorded contexts 0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 Cost per task, log scale GPT-5.2 Codex, ServiceX + Awkward Array + Hist · Combined current + historical contexts GPT-5 mini, ServiceX + ROOT RDataFrame · Combined recorded contexts

Source: Official LLM Physics Plotting Benchmark leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.2 Codex ServiceX + Awkward Array + Hist · Combined current + historical contexts 90.6% $0.10
GPT-5.2 Codex ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 90.6% $0.10
GPT-5.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6% $0.08
GPT-5.2 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 85.6% $0.08
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6% $0.03
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Current · 6261ce9ab00198a1 85.6% $0.03
GPT-5.6 Sol ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6% $0.33
GPT-5.6 Sol ServiceX + Awkward Array + Hist · Current · 6261ce9ab00198a1 85.6% $0.33
GPT-5 mini ServiceX + Awkward Array + Hist · Combined current + historical contexts 85% $0.04
GPT-5 mini ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 85% $0.04
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Combined current + historical contexts 84.4% $0.05
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Current · 6261ce9ab00198a1 84.4% $0.05
GLM-5 ServiceX + ROOT RDataFrame · Combined recorded contexts 83% $0.06
GLM-5 ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 83% $0.06
Gemini 2.5 Pro Jun 2025 ServiceX + Awkward Array + Hist · Combined current + historical contexts 82.2% $0.27
Gemini 2.5 Pro Jun 2025 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 82.2% $0.27
GLM-5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 81.1% $0.08
GLM-5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 81.1% $0.08
Kimi K2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 80.6% $0.06
Kimi K2.5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 80.6% $0.06
GPT-5 mini ServiceX + ROOT RDataFrame · Combined recorded contexts 77% $0.02
GPT-5 mini ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 77% $0.02
Claude Sonnet 4.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 74.4% $0.22
Claude Sonnet 4.5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 74.4% $0.22
GPT-5.2 ServiceX + ROOT RDataFrame · Combined recorded contexts 74% $0.07
GPT-5.2 ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 74% $0.07
Claude Opus 4.6 ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.9% $0.28
Claude Opus 4.6 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 73.9% $0.28
Gemini 3 Flash ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.3% $0.04
Gemini 3 Flash ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 73.3% $0.04
Kimi K2.5 ServiceX + ROOT RDataFrame · Combined recorded contexts 72% $0.05
Kimi K2.5 ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 72% $0.05
MiniMax-M2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 65% $0.05
MiniMax-M2.5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 65% $0.05
gpt-oss-120b ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6% $0.03
gpt-oss-120b ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 60.6% $0.03
grok-code-fast-1 ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6% $0.06
grok-code-fast-1 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 60.6% $0.06
DeepSeek-V3.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 53.9% $0.04
DeepSeek-V3.2 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 53.9% $0.04
gpt-oss-20b ServiceX + Awkward Array + Hist · Combined current + historical contexts 45% $0.02
gpt-oss-20b ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 45% $0.02
Llama 3.3 70B ServiceX + Awkward Array + Hist · Combined current + historical contexts 25% $0.03
Llama 3.3 70B ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 25% $0.03

The best score, 90.6%, costs $0.10 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-5.6 Luna at ServiceX + Awkward Array + Hist · Combined current + historical contexts comes within 5 points of the best for $0.03 a task, 3 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

OpenAI Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.01 $0.02 $0.05 $0.10 $0.20 $0.50 $1.00 Cost per task, log scale GPT-5.2 Codex, ServiceX + Awkward Array + Hist · Combined current + historical contexts GPT-5.6 Luna, ServiceX + Awkward Array + Hist · Combined current + historical contexts GPT-5 mini, ServiceX + ROOT RDataFrame · Combined recorded contexts 0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 Cost per task, log scale GPT-5.2 Codex, ServiceX + Awkward Array + Hist · Combined current + historical contexts GPT-5 mini, ServiceX + ROOT RDataFrame · Combined recorded contexts

Source: Official LLM Physics Plotting Benchmark leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.2 Codex ServiceX + Awkward Array + Hist · Combined current + historical contexts 90.6% $0.10
GPT-5.2 Codex ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 90.6% $0.10
GPT-5.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6% $0.08
GPT-5.2 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 85.6% $0.08
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6% $0.03
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Current · 6261ce9ab00198a1 85.6% $0.03
GPT-5.6 Sol ServiceX + Awkward Array + Hist · Combined current + historical contexts 85.6% $0.33
GPT-5.6 Sol ServiceX + Awkward Array + Hist · Current · 6261ce9ab00198a1 85.6% $0.33
GPT-5 mini ServiceX + Awkward Array + Hist · Combined current + historical contexts 85% $0.04
GPT-5 mini ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 85% $0.04
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Combined current + historical contexts 84.4% $0.05
GPT-5.6 Luna ServiceX + Awkward Array + Hist · Current · 6261ce9ab00198a1 84.4% $0.05
GLM-5 ServiceX + ROOT RDataFrame · Combined recorded contexts 83% $0.06
GLM-5 ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 83% $0.06
Gemini 2.5 Pro Jun 2025 ServiceX + Awkward Array + Hist · Combined current + historical contexts 82.2% $0.27
Gemini 2.5 Pro Jun 2025 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 82.2% $0.27
GLM-5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 81.1% $0.08
GLM-5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 81.1% $0.08
Kimi K2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 80.6% $0.06
Kimi K2.5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 80.6% $0.06
GPT-5 mini ServiceX + ROOT RDataFrame · Combined recorded contexts 77% $0.02
GPT-5 mini ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 77% $0.02
Claude Sonnet 4.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 74.4% $0.22
Claude Sonnet 4.5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 74.4% $0.22
GPT-5.2 ServiceX + ROOT RDataFrame · Combined recorded contexts 74% $0.07
GPT-5.2 ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 74% $0.07
Claude Opus 4.6 ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.9% $0.28
Claude Opus 4.6 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 73.9% $0.28
Gemini 3 Flash ServiceX + Awkward Array + Hist · Combined current + historical contexts 73.3% $0.04
Gemini 3 Flash ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 73.3% $0.04
Kimi K2.5 ServiceX + ROOT RDataFrame · Combined recorded contexts 72% $0.05
Kimi K2.5 ServiceX + ROOT RDataFrame · Historical import · legacy-b55696a333ff 72% $0.05
MiniMax-M2.5 ServiceX + Awkward Array + Hist · Combined current + historical contexts 65% $0.05
MiniMax-M2.5 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 65% $0.05
gpt-oss-120b ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6% $0.03
gpt-oss-120b ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 60.6% $0.03
grok-code-fast-1 ServiceX + Awkward Array + Hist · Combined current + historical contexts 60.6% $0.06
grok-code-fast-1 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 60.6% $0.06
DeepSeek-V3.2 ServiceX + Awkward Array + Hist · Combined current + historical contexts 53.9% $0.04
DeepSeek-V3.2 ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 53.9% $0.04
gpt-oss-20b ServiceX + Awkward Array + Hist · Combined current + historical contexts 45% $0.02
gpt-oss-20b ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 45% $0.02
Llama 3.3 70B ServiceX + Awkward Array + Hist · Combined current + historical contexts 25% $0.03
Llama 3.3 70B ServiceX + Awkward Array + Hist · Historical import · legacy-d63e64dde22b 25% $0.03

3 models have held number one on the official leaderboard since its release in September 2025

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

OpenAI
Own leaderboard Passed 85% in Sep 2025, by GPT-5 mini

Source: Official LLM Physics Plotting Benchmark leaderboard

GPT-5 mini on the official leaderboard passed 85% in August 2025

Models released before it came out already scored that high. 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 2026 Near the ceiling (85%) GPT-5.2 Codex GPT-5.2 GPT-5 mini Llama 3.3 70B 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) GPT-5.2 Codex Gemini 2.5 Pro

Source: Official LLM Physics Plotting Benchmark 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

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (19 entries).
  • Pass Models scored 17 models we track, plus 2 other entries on its leaderboard.
  • Pass Newest score The newest is from July 2026, for GPT-5.6 Luna.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.6 Sol (#14).
  • Note Headroom Saturated. The best is 90.6%, past 85% since August 2025.
Leaderboard import 3 pass
  • Pass Importer We read the board on gordonwatts.github.io: 44 rows for models we track and 2 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