Benchmark Almanac

Blueprint-Bench 2

Gemini 4 Carbon 58.6% leads since Oct 2026 In use

Agents turn about 20 interior photos of each of 50 apartments into 2D floor plans. Scores compare room connectivity with ground truth, normalized against random.

The leader now ties the human baseline, which was measured on only 12 apartments.

Made by Andon Labs Released May 2026 Spatial reasoning and counting Tasks kept private Last in a launch post Oct 2026 Last model added Oct 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
Open-ended Free-form text, documents, designs or media judged as a whole.
Leaderboard andonlabs.com
35
Models scored on its own leaderboard
3
Launch posts list it, from 2 labs
2
System cards report it, from 2 labs

Gemini 4 Carbon leads with 58.6%

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

35 models. Scroll sideways for the rest.

  1. 58.6 Gemini 4 Carbon
  2. 54.4 Gemini 4 Argon
  3. 51.2 Claude Opus 5.5
  4. 49.7 GPT-6 Astra
  5. 42.1 GPT-6.1 Terra
  6. 41.9 Claude Fable 5.1
  7. 39.3 Claude Sonnet 5.5
  8. 38.6 Claude Fable 5
  9. 38.6 Gemini 3.8 Flash
  10. 36.9 GPT-6 Sol
  11. 36.2 GPT-5.5
  12. 33.6 Gemini 3.5 Flash
  13. 33.6 GPT-5.6 Sol
  14. 33.2 Grok 4.6
  15. 33.0 Claude Haiku 5.5
  16. 32.5 Grok 4.7
  17. 31.2 Gemini 3.6 Flash
  18. 31.2 GPT-6 Luna
  19. 30.8 GPT-5.6 Terra
  20. 30.4 Claude Opus 5
  21. 29.5 Kimi K3
  22. 27.3 Grok 4.5
  23. 27.1 GPT-5.4
  24. 26.5 Gemini 3.1 Pro
  25. 24.9 Claude Sonnet 5
  26. 24.5 Claude Opus 4.7
  27. 22.6 GPT-5.6 Luna
  28. 14.5 Claude Opus 4.8
  29. 6.70 Claude Sonnet 4.6
  30. 3.90 Kimi K2.6
  31. 0.0 Claude Haiku 4.5
  32. 0.0 Gemini 3 Flash
  33. 0.0 Grok 4.20
  34. 0.0 Gemini Robotics-ER 1.6
  35. 0.0 Grok 4.3
  1. 1 Gemini 4 Carbon 58.6%
  2. 2 Gemini 4 Argon 54.4%
  3. 3 Claude Opus 5.5 51.2%
  4. 4 GPT-6 Astra 49.7%
  5. 5 GPT-6.1 Terra 42.1%
  6. 6 Claude Fable 5.1 41.9%
  7. 7 Claude Sonnet 5.5 39.3%
  8. 8 Claude Fable 5 38.6%
  9. 9 Gemini 3.8 Flash 38.6%
  10. 10 GPT-6 Sol 36.9%
  11. 11 GPT-5.5 36.2%
  12. 12 Gemini 3.5 Flash 33.6%
  13. 13 GPT-5.6 Sol 33.6%
  14. 14 Grok 4.6 33.2%
  15. 15 Claude Haiku 5.5 33%
  16. 16 Grok 4.7 32.5%
  17. 17 Gemini 3.6 Flash 31.2%
  18. 18 GPT-6 Luna 31.2%
  19. 19 GPT-5.6 Terra 30.8%
  20. 20 Claude Opus 5 30.4%
  21. 21 Kimi K3 29.5%
  22. 22 Grok 4.5 27.3%
  23. 23 GPT-5.4 27.1%
  24. 24 Gemini 3.1 Pro 26.5%
  25. 25 Claude Sonnet 5 24.9%
  26. 26 Claude Opus 4.7 24.5%
  27. 27 GPT-5.6 Luna 22.6%
  28. 28 Claude Opus 4.8 14.5%
  29. 29 Claude Sonnet 4.6 6.70%
  30. 30 Kimi K2.6 3.90%
  31. 31 Claude Haiku 4.5 0%
  32. 32 Gemini 3 Flash 0%
  33. 33 Grok 4.20 reasoning 0%
  34. 34 Gemini Robotics-ER 1.6 0%
  35. 35 Grok 4.3 0%
#ModelScoreSettingDateSource
1 Gemini 4 Carbon 58.6% Oct 2026 Official leaderboard
2 Gemini 4 Argon 54.4% Sep 2026 Official leaderboard
3 Claude Opus 5.5 51.2% Sep 2026 Official leaderboard
4 GPT-6 Astra 49.7% Sep 2026 Official leaderboard
5 GPT-6.1 Terra 42.1% Oct 2026 Official leaderboard
6 Claude Fable 5.1 41.9% Sep 2026 Official leaderboard
7 Claude Sonnet 5.5 39.3% Sep 2026 Official leaderboard
8 Claude Fable 5 38.6% Jun 2026 Official leaderboard
9 Gemini 3.8 Flash 38.6% Sep 2026 Official leaderboard
10 GPT-6 Sol 36.9% Sep 2026 Official leaderboard
11 GPT-5.5 36.2% Apr 2026 Official leaderboard
12 Gemini 3.5 Flash 33.6% May 2026 Official leaderboard
13 GPT-5.6 Sol 33.6% Jul 2026 Official leaderboard
14 Grok 4.6 33.2% Aug 2026 Official leaderboard
15 Claude Haiku 5.5 33% Oct 2026 Official leaderboard
16 Grok 4.7 32.5% Sep 2026 Official leaderboard
17 Gemini 3.6 Flash 31.2% Jul 2026 Official leaderboard
18 GPT-6 Luna 31.2% Sep 2026 Official leaderboard
19 GPT-5.6 Terra 30.8% Jul 2026 Official leaderboard
20 Claude Opus 5 30.4% Jul 2026 Official leaderboard
21 Kimi K3 29.5% Jul 2026 Official leaderboard
22 Grok 4.5 27.3% Jul 2026 Official leaderboard
23 GPT-5.4 27.1% Mar 2026 Official leaderboard
24 Gemini 3.1 Pro 26.5% Feb 2026 Official leaderboard
25 Claude Sonnet 5 24.9% Jun 2026 Official leaderboard
26 Claude Opus 4.7 24.5% Apr 2026 Official leaderboard
27 GPT-5.6 Luna 22.6% Jul 2026 Official leaderboard
28 Claude Opus 4.8 14.5% May 2026 Official leaderboard
29 Claude Sonnet 4.6 6.70% Feb 2026 Official leaderboard
30 Kimi K2.6 3.90% Apr 2026 Official leaderboard
31 Claude Haiku 4.5 0% Oct 2025 Official leaderboard
32 Gemini 3 Flash 0% Dec 2025 Official leaderboard
33 Grok 4.20 0% reasoning Feb 2026 Official leaderboard
34 Gemini Robotics-ER 1.6 0% Apr 2026 Official leaderboard
35 Grok 4.3 0% Apr 2026 Official leaderboard

Epoch AI has scores for 35 models. Labs report scores for 3 models in their cards. We plot every source in the charts below.

7 models have held number one since its release in May 2026

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard, Epoch AI’s runs and what labs claim in their cards.

OpenAI Anthropic Google DeepMind
Own leaderboard 7 leaders
Gemini 4 Carbon 58.6%
Epoch AI runs 7 leaders
Gemini 4 Carbon 58.6%
Lab claims 3 leaders
Gemini 4 Carbon 58.6%

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 Run or collected by Epoch AI Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Gemini 4 Carbon Gemini 4 Argon GPT-6 Astra Claude Fable 5.1 GPT-5.5 GPT-5.4 Claude Sonnet 4.6 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Gemini 4 Carbon Claude Opus 5.5

2 labs put it in a launch post in the last six months

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

Google DeepMind
Anthropic
DateLaunch postScoreAs the post writes it
28 Oct 2026 Gemini 4 Carbon 58.6 Blueprint-Bench 2 (Similarity score)
9 Jun 2026 Claude Fable 5 38.6 Spatial reasoning Blueprint-Bench 2
19 May 2026 Gemini 3.5 Flash 33.6 Blueprint-Bench 2 (Normalized score)

2 system cards report it, from 2 labs

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
9 Jun 2026 Claude Fable 5 system card 38.6 Blueprint-Bench 2, normalized composite score on 50 apartments, run by Andon Labs, max effort with coding tools (Table 8.1.A, 8.16.2) Capabilities
19 May 2026 Gemini 3.5 Flash model card 33.6 Blueprint-Bench 2, normalized score (results table, p.4) Capabilities

Versions

Replaces Blueprint-Bench

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%