Benchmark Almanac

ClockBench

Gemini 4 Argon 83.3% leads since Sep 2026 Not in model cards

Models read analog clock faces and answer questions about the time shown. Most clocks stay private to avoid training leakage.

180 clocks with 720 questions, 10 clocks public. Humans score about 90.7% on average. More than 30 models on the board, led by Gemini 4 Argon at 83.3% in 2026-10.

Made by Alek Safar Released Sep 2025 Spatial reasoning and counting Public tasks plus a private test set Last model added Sep 2026 Scores refreshed 2 Nov 2026
Level
Everyday Most adults, without training
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Short answer A number, word or short string matched to a key.
aleksafar/clockbench 34 stars Paper Leaderboard clockbench.ai
39
Entries on its own leaderboard, 37 models we track

Gemini 4 Argon leads with 83.3%

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 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.

39 entries. Scroll sideways for the rest.

  1. 83.3 Gemini 4 Argon
  2. 77.2 Claude Opus 5.5
  3. 75.0 GPT-6 Astra
  4. 74.4 GPT-6.1 Sol
  5. 68.3 GPT-5.6 Sol
  6. 66.7 Gemini 3.8 Flash
  7. 65.5 Claude Opus 5
  8. 59.4 Claude Fable 5.1
  9. 53.3 Gemini 3.7 Flash
  10. 50.6 GPT-5.4
  11. 46.1 GPT-5.5
  12. 42.2 Muse Spark 1.3
  13. 39.4 Qwen3-VL-235B-A22B-Instruct
  14. 37.2 Claude Fable 5
  15. 34.4 Gemini 3.5 Flash
  16. 32.2 Gemini 3.1 Pro
  17. 31.1 Gemini 3 Pro
  18. 21.7 Grok 4.5
  19. 20.0 Grok 4.6
  20. 18.9 Gemini 2.5 Pro
  21. 15.6 GPT-5.2
  22. 15.0 gemini-robotics-er-1.5-preview
  23. 15.0 Claude Opus 4.7
  24. 14.4 Qwen 3-VL 235B Thinking
  25. 14.4 o3-pro
  26. 12.2 o3
  27. 11.1 GPT-5
  28. 11.1 Gemini 2.5 Flash
  29. 11.1 GPT-5 Pro
  30. 10.6 Mistral Medium 3.1
  31. 10.0 Claude Opus 4.6
  32. 8.90 Claude Opus 4.1
  33. 8.90 GPT-5 mini
  34. 7.80 Claude Sonnet 4.5
  35. 6.70 GTP-4o
  36. 6.70 Claude Sonnet 4
  37. 6.10 Qwen2.5-VL-72B
  38. 3.90 GPT-5 nano
  39. 3.30 Grok 4 Fast
  1. 1 Gemini 4 Argon high 83.3%
  2. 2 Claude Opus 5.5 max 77.2%
  3. 3 GPT-6 Astra max 75%
  4. 4 GPT-6.1 Sol max 74.4%
  5. 5 GPT-5.6 Sol max 68.3%
  6. 6 Gemini 3.8 Flash 66.7%
  7. 7 Claude Opus 5 max 65.5%
  8. 8 Claude Fable 5.1 max 59.4%
  9. 9 Gemini 3.7 Flash 53.3%
  10. 10 GPT-5.4 high 50.6%
  11. 11 GPT-5.5 high 46.1%
  12. 12 Muse Spark 1.3 42.2%
  13. 13 Qwen3-VL-235B-A22B-Instruct 39.4%
  14. 14 Claude Fable 5 37.2%
  15. 15 Gemini 3.5 Flash 34.4%
  16. 16 Gemini 3.1 Pro 32.2%
  17. 17 Gemini 3 Pro 31.1%
  18. 18 Grok 4.5 21.7%
  19. 19 Grok 4.6 20%
  20. 20 Gemini 2.5 Pro Jun 2025 18.9%
  21. 21 GPT-5.2 high 15.6%
  22. 22 gemini-robotics-er-1.5-preview 15%
  23. 23 Claude Opus 4.7 15%
  24. 24 Qwen 3-VL 235B Thinking 14.4%
  25. 25 o3-pro 14.4%
  26. 26 o3 high 12.2%
  27. 27 GPT-5 high 11.1%
  28. 28 Gemini 2.5 Flash Sep 2025 11.1%
  29. 29 GPT-5 Pro 11.1%
  30. 30 Mistral Medium 3.1 10.6%
  31. 31 Claude Opus 4.6 10%
  32. 32 Claude Opus 4.1 8.90%
  33. 33 GPT-5 mini 8.90%
  34. 34 Claude Sonnet 4.5 7.80%
  35. 35 GTP-4o 6.70%
  36. 36 Claude Sonnet 4 6.70%
  37. 37 Qwen2.5-VL-72B 6.10%
  38. 38 GPT-5 nano 3.90%
  39. 39 Grok 4 Fast 3.30%
#EntryScoreSettingDateSource
1 Gemini 4 Argon 83.3% high Sep 2026 Official leaderboard
2 Claude Opus 5.5 77.2% max Sep 2026 Official leaderboard
3 GPT-6 Astra 75% max Sep 2026 Official leaderboard
4 GPT-6.1 Sol 74.4% max Sep 2026 Official leaderboard
5 GPT-5.6 Sol 68.3% max Jul 2026 Official leaderboard
6 Gemini 3.8 Flash 66.7% Sep 2026 Official leaderboard
7 Claude Opus 5 65.5% max Jul 2026 Official leaderboard
8 Claude Fable 5.1 59.4% max Sep 2026 Official leaderboard
9 Gemini 3.7 Flash 53.3% Aug 2026 Official leaderboard
10 GPT-5.4 50.6% high Mar 2026 Official leaderboard
11 GPT-5.5 46.1% high Apr 2026 Official leaderboard
12 Muse Spark 1.3 42.2% Sep 2026 Official leaderboard
13 Qwen3-VL-235B-A22B-Instruct 39.4% Sep 2025 Official leaderboard
14 Claude Fable 5 37.2% Jun 2026 Official leaderboard
15 Gemini 3.5 Flash 34.4% May 2026 Official leaderboard
16 Gemini 3.1 Pro 32.2% Feb 2026 Official leaderboard
17 Gemini 3 Pro 31.1% Nov 2025 Official leaderboard
18 Grok 4.5 21.7% Jul 2026 Official leaderboard
19 Grok 4.6 20% Aug 2026 Official leaderboard
20 Gemini 2.5 Pro Jun 2025 18.9% Jun 2025 Official leaderboard
21 GPT-5.2 15.6% high Dec 2025 Official leaderboard
22 gemini-robotics-er-1.5-preview 15% Sep 2025 Official leaderboard
23 Claude Opus 4.7 15% Apr 2026 Official leaderboard
24 Qwen 3-VL 235B Thinking 14.4% Official leaderboard
25 o3-pro 14.4% Jun 2025 Official leaderboard
26 o3 12.2% high Apr 2025 Official leaderboard
27 GPT-5 11.1% high Aug 2025 Official leaderboard
28 Gemini 2.5 Flash Sep 2025 11.1% Sep 2025 Official leaderboard
29 GPT-5 Pro 11.1% Oct 2025 Official leaderboard
30 Mistral Medium 3.1 10.6% Aug 2025 Official leaderboard
31 Claude Opus 4.6 10% Feb 2026 Official leaderboard
32 Claude Opus 4.1 8.90% Aug 2025 Official leaderboard
33 GPT-5 mini 8.90% Aug 2025 Official leaderboard
34 Claude Sonnet 4.5 7.80% Sep 2025 Official leaderboard
35 GTP-4o 6.70% Official leaderboard
36 Claude Sonnet 4 6.70% May 2025 Official leaderboard
37 Qwen2.5-VL-72B 6.10% Jan 2025 Official leaderboard
38 GPT-5 nano 3.90% Aug 2025 Official leaderboard
39 Grok 4 Fast 3.30% Sep 2025 Official leaderboard

7 models have held number one since its release in Sep 2025

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

Google DeepMind Qwen OpenAI Anthropic
Own leaderboard 7 leaders
Gemini 4 Argon 83.3%

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Gemini 4 Argon Claude Opus 5.5 GPT-5.6 Sol GPT-5.4 Qwen3-VL-235B-A22B-Instruct Gemini 2.5 Pro o3 0% 20% 40% 60% 80% 100% 2025 2026 Near the ceiling (85%) Gemini 4 Argon GPT-6 Astra

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Vision

All 234 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. GPT-5.1 85.4%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. Kimi K3.1 Pro 84.1%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • ChartQA In use Models answer human-written and generated questions about chart images that need visual and arithmetic reasoning. Scored by relaxed accuracy with a 5 percent numeric tolerance. Command A Vision 90.9%
  • DocVQA In use Models answer 50,000 questions about 12,000 scanned document images by reading their text and layout. Scoring is ANLS, a string similarity to reference answers. Seed1.5-VL 96.9%