Benchmark Almanac

Chartography

GPT-6.2 Astra 90.7% leads since November 2026
GPT-6.2 Astra 90.7% leads since November 2026
In use

The official leaderboard hasn’t scored a model released after Gemini 4 Neon (November 2026). Labs report their own numbers for one newer model: Inkling 2 Mini at 22.1%.

100 tasks on specialized professional charts (Kaplan-Meier, candlestick, Sankey, Bode). Answers graded against expert references.

Made by
Surge AI
Released
August 2026
Kind
Documents, charts and text in images
Level
Expert
Scored
Checked automatically
Answers
Short answer
Measures
Capability
Language
English
In and out
Text, Images → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Occupations
Business and financial operations, Architecture and engineering, Life, physical and social science, Healthcare practitioners and technical
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last in a launch post
November 2026
Last model added
November 2026
Scores refreshed
November 20th, 2026

In the authors’ words

Professionals across medicine, engineering, finance, manufacturing, and the sciences often make consequential decisions from charts. Existing chart benchmarks do not sufficiently measure this ability: they are dominated by bar, line, and pie formats, rely on shorter reasoning chains, and are nearing saturation, with frontier models already scoring 80-90%. We introduce Chartography, a benchmark of 100 tasks that pair charts drawn from professional practice, in domain-specific formats that standard chart benchmarks rarely include, with questions written by professionals who read these charts for a living and independently verified by three additional experts. In an evaluation of 30 frontier-model configurations (20 scored trials per task), the best configuration reaches only 45.0% mean pass@1; the remainder span 9.0-39.5%. Failures concentrate in visual perception: models can miss nuanced features, misread values along sparsely labeled axes, mishandle projected 3D geometry, and violate domain conventions encoded in the chart. We release all tasks, images, provenance metadata, and evaluation code.

Abstract of Chartography: A Benchmark for Professional Chart Understanding (arXiv:2608.10677, CC0)
56
Models scored on its own leaderboard
16
Launch posts list it, from 7 labs
11
System cards report it, from 4 labs

GPT-6.2 Astra leads with 90.7%

Best score per model 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.

GPT-6.2 Astra leads with 90.7%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

  1. 1 GPT-6.2 Astra max 90.7%
  2. 2 Claude Opus 5.5 adaptive/max 89.6%
  3. 3 GPT-6 Astra max 88.6%
  4. 4 GPT-6.2 Sol max 87.7%
  5. 5 Gemini 4 Hydrogen xhigh 87.2%
  6. 6 Gemini 4 Neon high 86.6%
  7. 7 Gemini 4 Carbon high 85.5%
  8. 8 Muse Spark 1.3 max 84.6%
  9. 9 Claude Opus 5 adaptive/max 84.4%
  10. 10 Gemini 3.8 Flash high 81.7%
  11. 11 Claude Fable 5.2 adaptive/max 80.8%
  12. 12 Claude Fable 5.1 adaptive/max 79.6%
  13. 13 GPT-6 Sol max 79.4%
  14. 14 Muse Ember max 79.3%
  15. 15 Claude Fable 5 adaptive/max 79%
  16. 16 GPT-5.6 Sol max 78.6%
  17. 17 DeepSeek-V4-Flash max 77.1%
  18. 18 Inkling 2 Small xhigh 75.6%
  19. 19 Qwen4-Max xhigh 75.3%
  20. 20 Qwen 3.8 Max xhigh 75.2%
  21. 21 Gemini 3.7 Flash high 74.9%
  22. 22 Grok 4.7 xhigh 74.6%
  23. 23 GLM-5.3-Flash max 72.7%
  24. 24 GPT-5.6 Luna max 72.2%
  25. 25 GPT-6 Luna max 68%
  26. 26 Muse Spark 1.2 xhigh 66.7%
  27. 27 Kimi K3 max 66.4%
  28. 28 GLM-5.3 max 64.2%
  29. 29 Claude Sonnet 5 adaptive/max 64%
  30. 30 Tencent Hy4 preview high 59.4%
  31. 31 GPT-5.6 Terra xhigh 58.8%
  32. 32 Gemini 3.1 Pro high 54.2%
  33. 33 Tencent Hy4 high 49.2%
  34. 34 Grok 4.6 xhigh 44.4%
  35. 35 MiniMax-M3.1-Flash max 39.8%
  36. 36 Gemini 3.5 Flash Pareto chart; effort unspecified 35.9%
  37. 37 Gemini 3.6 Flash Pareto chart; effort unspecified 34%
  38. 38 Kimi K2.7 Code max 31.5%
  39. 39 GPT-5.5 Pareto chart; xhigh 31%
  40. 40 GPT-5.4 Pareto chart; xhigh 29.6%
  41. 41 Muse Spark 1.1 Pareto chart; xhigh 24.4%
  42. 42 DeepSeek-V4-Pro max 21.1%
  43. 43 Tencent Hy3 high 20.3%
  44. 44 Grok 4.5 Pareto chart; effort unspecified 17.3%
  45. 45 Claude Opus 4.7 Pareto chart; max 16.5%
  46. 46 Qwen3.5-Plus Pareto chart; effort unspecified 15.9%
  47. 47 Claude Opus 4.8 Pareto chart; max 15.9%
  48. 48 Qwen3.7-Plus Pareto chart; effort unspecified 15.8%
  49. 49 Muse Glimmer xhigh 15.4%
  50. 50 Inkling high 15%
  51. 51 Grok 4.3 Pareto chart; high 12.9%
  52. 52 Kimi K2.5 Pareto chart; effort unspecified 12.6%
  53. 53 Kimi K2.6 Pareto chart; effort unspecified 12.2%
  54. 54 Gemini 3.5 Flash-Lite Pareto chart; effort unspecified 10.2%
  55. 55 Mistral Large 3 Pareto chart; effort unspecified 9.00%
  56. 56 Nemotron 3 Ultra 6.80%

Source: Official Chartography leaderboard

  1. 1 GPT-6.2 Astra max 90.7%
  2. 2 Kimi K3.1 Pro 90.2%
  3. 3 Claude Opus 5.5 adaptive/max 89.6%
  4. 4 GPT-6 Astra max 88.6%
  5. 5 GPT-6.2 Sol max 87.7%
  6. 6 Gemini 4 Hydrogen xhigh 87.2%
  7. 7 Gemini 4 Neon high 86.6%
  8. 8 Gemini 4 Carbon high 85.5%
  9. 9 Muse Spark 1.3 max 84.6%
  10. 10 Claude Opus 5 adaptive/max 84.4%
  11. 11 DeepSeek V4.1 Flash 1017 83%
  12. 12 Gemini 3.8 Flash high 81.7%
  13. 13 Claude Fable 5.2 adaptive/max 80.8%
  14. 14 Claude Fable 5.1 adaptive/max 79.6%
  15. 15 GPT-6 Sol max 79.4%
  16. 16 Muse Ember max 79.3%
  17. 17 Claude Fable 5 adaptive/max 79%
  18. 18 DeepSeek V4.1 Flash 78.9%
  19. 19 GPT-5.6 Sol max 78.6%
  20. 20 GLM-5.3-Flash 78%
  21. 21 DeepSeek-V4-Flash max 77.1%
  22. 22 Inkling 2 Small xhigh 75.6%
  23. 23 Qwen4-Max xhigh 75.3%
  24. 24 Qwen 3.8 Max xhigh 75.2%
  25. 25 Gemini 3.7 Flash high 74.9%
  26. 26 Grok 4.7 xhigh 74.6%
  27. 27 GPT-5.6 Luna max 72.2%
  28. 28 Gemini 4 Argon 71.6%
  29. 29 GPT-6 Luna max 68%
  30. 30 Muse Spark 1.2 xhigh 66.7%
  31. 31 Kimi K3 max 66.4%
  32. 32 DeepSeek-V4-Flash-Vision-Exp 64.3%
  33. 33 GLM-5.3 max 64.2%
  34. 34 Claude Sonnet 5 adaptive/max 64%
  35. 35 Claude Sonnet 5.5 61.6%
  36. 36 Tencent Hy4 preview high 59.4%
  37. 37 GPT-5.6 Terra xhigh 58.8%
  38. 38 Gemini 3.1 Pro high 54.2%
  39. 39 Tencent Hy4 high 49.2%
  40. 40 Claude Haiku 5.5 46.4%
  41. 41 Grok 4.6 xhigh 44.4%
  42. 42 MiniMax-M3.1-Flash max 39.8%
  43. 43 Gemini 3.5 Flash Pareto chart; effort unspecified 35.9%
  44. 44 Gemini 3.6 Flash Pareto chart; effort unspecified 34%
  45. 45 Kimi K2.7 Code max 31.5%
  46. 46 GPT-5.5 Pareto chart; xhigh 31%
  47. 47 GPT-5.4 Pareto chart; xhigh 29.6%
  48. 48 Muse Spark 1.1 Pareto chart; xhigh 24.4%
  49. 49 Inkling 2 Mini 22.1%
  50. 50 DeepSeek-V4-Pro max 21.1%
  51. 51 Tencent Hy3 high 20.3%
  52. 52 Grok 4.5 Pareto chart; effort unspecified 17.3%
  53. 53 Claude Opus 4.7 Pareto chart; max 16.5%
  54. 54 Qwen3.5-Plus Pareto chart; effort unspecified 15.9%
  55. 55 Claude Opus 4.8 Pareto chart; max 15.9%
  56. 56 Qwen3.7-Plus Pareto chart; effort unspecified 15.8%
  57. 57 Muse Glimmer xhigh 15.4%
  58. 58 Inkling high 15%
  59. 59 Grok 4.3 Pareto chart; high 12.9%
  60. 60 Kimi K2.5 Pareto chart; effort unspecified 12.6%
  61. 61 Kimi K2.6 Pareto chart; effort unspecified 12.2%
  62. 62 Gemini 3.5 Flash-Lite Pareto chart; effort unspecified 10.2%
  63. 63 Mistral Large 3 Pareto chart; effort unspecified 9.00%
  64. 64 Nemotron 3 Ultra 6.80%

Sources: Official Chartography leaderboard and labs’ model cards

#ModelScoreSettingDateSource
1 GPT-6.2 Astra 90.7% max Nov 2026 Official leaderboard
2 Claude Opus 5.5 89.6% adaptive/max Sep 2026 Official leaderboard
3 GPT-6 Astra 88.6% max Sep 2026 Official leaderboard
4 GPT-6.2 Sol 87.7% max Nov 2026 Official leaderboard
5 Gemini 4 Hydrogen 87.2% xhigh Nov 2026 Official leaderboard
6 Gemini 4 Neon 86.6% high Nov 2026 Official leaderboard
7 Gemini 4 Carbon 85.5% high Oct 2026 Official leaderboard
8 Muse Spark 1.3 84.6% max Sep 2026 Official leaderboard
9 Claude Opus 5 84.4% adaptive/max Jul 2026 Official leaderboard
10 Gemini 3.8 Flash 81.7% high Sep 2026 Official leaderboard
11 Claude Fable 5.2 80.8% adaptive/max Nov 2026 Official leaderboard
12 Claude Fable 5.1 79.6% adaptive/max Sep 2026 Official leaderboard
13 GPT-6 Sol 79.4% max Sep 2026 Official leaderboard
14 Muse Ember 79.3% max Nov 2026 Official leaderboard
15 Claude Fable 5 79% adaptive/max Jun 2026 Official leaderboard
16 GPT-5.6 Sol 78.6% max Jul 2026 Official leaderboard
17 DeepSeek-V4-Flash 77.1% max Apr 2026 Official leaderboard
18 Inkling 2 Small 75.6% xhigh Nov 2026 Official leaderboard
19 Qwen4-Max 75.3% xhigh Nov 2026 Official leaderboard
20 Qwen 3.8 Max 75.2% xhigh Aug 2026 Official leaderboard
21 Gemini 3.7 Flash 74.9% high Aug 2026 Official leaderboard
22 Grok 4.7 74.6% xhigh Sep 2026 Official leaderboard
23 GLM-5.3-Flash 72.7% max Aug 2026 Official leaderboard
24 GPT-5.6 Luna 72.2% max Jul 2026 Official leaderboard
25 GPT-6 Luna 68% max Sep 2026 Official leaderboard
26 Muse Spark 1.2 66.7% xhigh Aug 2026 Official leaderboard
27 Kimi K3 66.4% max Jul 2026 Official leaderboard
28 GLM-5.3 64.2% max Aug 2026 Official leaderboard
29 Claude Sonnet 5 64% adaptive/max Jun 2026 Official leaderboard
30 Tencent Hy4 preview 59.4% high Aug 2026 Official leaderboard
31 GPT-5.6 Terra 58.8% xhigh Jul 2026 Official leaderboard
32 Gemini 3.1 Pro 54.2% high Feb 2026 Official leaderboard
33 Tencent Hy4 49.2% high Nov 2026 Official leaderboard
34 Grok 4.6 44.4% xhigh Aug 2026 Official leaderboard
35 MiniMax-M3.1-Flash 39.8% max Oct 2026 Official leaderboard
36 Gemini 3.5 Flash 35.9% Pareto chart; effort unspecified May 2026 Official leaderboard
37 Gemini 3.6 Flash 34% Pareto chart; effort unspecified Jul 2026 Official leaderboard
38 Kimi K2.7 Code 31.5% max Jun 2026 Official leaderboard
39 GPT-5.5 31% Pareto chart; xhigh Apr 2026 Official leaderboard
40 GPT-5.4 29.6% Pareto chart; xhigh Mar 2026 Official leaderboard
41 Muse Spark 1.1 24.4% Pareto chart; xhigh Jul 2026 Official leaderboard
42 DeepSeek-V4-Pro 21.1% max Apr 2026 Official leaderboard
43 Tencent Hy3 20.3% high Jul 2026 Official leaderboard
44 Grok 4.5 17.3% Pareto chart; effort unspecified Jul 2026 Official leaderboard
45 Claude Opus 4.7 16.5% Pareto chart; max Apr 2026 Official leaderboard
46 Qwen3.5-Plus 15.9% Pareto chart; effort unspecified Feb 2026 Official leaderboard
47 Claude Opus 4.8 15.9% Pareto chart; max May 2026 Official leaderboard
48 Qwen3.7-Plus 15.8% Pareto chart; effort unspecified Jun 2026 Official leaderboard
49 Muse Glimmer 15.4% xhigh Aug 2026 Official leaderboard
50 Inkling 15% high Jul 2026 Official leaderboard
51 Grok 4.3 12.9% Pareto chart; high Apr 2026 Official leaderboard
52 Kimi K2.5 12.6% Pareto chart; effort unspecified Jan 2026 Official leaderboard
53 Kimi K2.6 12.2% Pareto chart; effort unspecified Apr 2026 Official leaderboard
54 Gemini 3.5 Flash-Lite 10.2% Pareto chart; effort unspecified Jul 2026 Official leaderboard
55 Mistral Large 3 9.00% Pareto chart; effort unspecified Dec 2025 Official leaderboard
56 Nemotron 3 Ultra 6.80% Jun 2026 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-6.2 Astra 90.7% max Nov 2026 Official leaderboard
2 Kimi K3.1 Pro 90.2% Oct 2026 Lab model card
3 Claude Opus 5.5 89.6% adaptive/max Sep 2026 Official leaderboard
4 GPT-6 Astra 88.6% max Sep 2026 Official leaderboard
5 GPT-6.2 Sol 87.7% max Nov 2026 Official leaderboard
6 Gemini 4 Hydrogen 87.2% xhigh Nov 2026 Official leaderboard
7 Gemini 4 Neon 86.6% high Nov 2026 Official leaderboard
8 Gemini 4 Carbon 85.5% high Oct 2026 Official leaderboard
9 Muse Spark 1.3 84.6% max Sep 2026 Official leaderboard
10 Claude Opus 5 84.4% adaptive/max Jul 2026 Official leaderboard
11 DeepSeek V4.1 Flash 1017 83% Oct 2026 Lab model card
12 Gemini 3.8 Flash 81.7% high Sep 2026 Official leaderboard
13 Claude Fable 5.2 80.8% adaptive/max Nov 2026 Official leaderboard
14 Claude Fable 5.1 79.6% adaptive/max Sep 2026 Official leaderboard
15 GPT-6 Sol 79.4% max Sep 2026 Official leaderboard
16 Muse Ember 79.3% max Nov 2026 Official leaderboard
17 Claude Fable 5 79% adaptive/max Jun 2026 Official leaderboard
18 DeepSeek V4.1 Flash 78.9% Sep 2026 Lab model card
19 GPT-5.6 Sol 78.6% max Jul 2026 Official leaderboard
20 GLM-5.3-Flash 78% Aug 2026 Lab model card
21 DeepSeek-V4-Flash 77.1% max Apr 2026 Official leaderboard
22 Inkling 2 Small 75.6% xhigh Nov 2026 Official leaderboard
23 Qwen4-Max 75.3% xhigh Nov 2026 Official leaderboard
24 Qwen 3.8 Max 75.2% xhigh Aug 2026 Official leaderboard
25 Gemini 3.7 Flash 74.9% high Aug 2026 Official leaderboard
26 Grok 4.7 74.6% xhigh Sep 2026 Official leaderboard
27 GPT-5.6 Luna 72.2% max Jul 2026 Official leaderboard
28 Gemini 4 Argon 71.6% Oct 2026 Lab model card
29 GPT-6 Luna 68% max Sep 2026 Official leaderboard
30 Muse Spark 1.2 66.7% xhigh Aug 2026 Official leaderboard
31 Kimi K3 66.4% max Jul 2026 Official leaderboard
32 DeepSeek-V4-Flash-Vision-Exp 64.3% Aug 2026 Lab model card
33 GLM-5.3 64.2% max Aug 2026 Official leaderboard
34 Claude Sonnet 5 64% adaptive/max Jun 2026 Official leaderboard
35 Claude Sonnet 5.5 61.6% Sep 2026 Lab model card
36 Tencent Hy4 preview 59.4% high Aug 2026 Official leaderboard
37 GPT-5.6 Terra 58.8% xhigh Jul 2026 Official leaderboard
38 Gemini 3.1 Pro 54.2% high Feb 2026 Official leaderboard
39 Tencent Hy4 49.2% high Nov 2026 Official leaderboard
40 Claude Haiku 5.5 46.4% Oct 2026 Lab model card
41 Grok 4.6 44.4% xhigh Aug 2026 Official leaderboard
42 MiniMax-M3.1-Flash 39.8% max Oct 2026 Official leaderboard
43 Gemini 3.5 Flash 35.9% Pareto chart; effort unspecified May 2026 Official leaderboard
44 Gemini 3.6 Flash 34% Pareto chart; effort unspecified Jul 2026 Official leaderboard
45 Kimi K2.7 Code 31.5% max Jun 2026 Official leaderboard
46 GPT-5.5 31% Pareto chart; xhigh Apr 2026 Official leaderboard
47 GPT-5.4 29.6% Pareto chart; xhigh Mar 2026 Official leaderboard
48 Muse Spark 1.1 24.4% Pareto chart; xhigh Jul 2026 Official leaderboard
49 Inkling 2 Mini 22.1% Nov 2026 Lab model card
50 DeepSeek-V4-Pro 21.1% max Apr 2026 Official leaderboard
51 Tencent Hy3 20.3% high Jul 2026 Official leaderboard
52 Grok 4.5 17.3% Pareto chart; effort unspecified Jul 2026 Official leaderboard
53 Claude Opus 4.7 16.5% Pareto chart; max Apr 2026 Official leaderboard
54 Qwen3.5-Plus 15.9% Pareto chart; effort unspecified Feb 2026 Official leaderboard
55 Claude Opus 4.8 15.9% Pareto chart; max May 2026 Official leaderboard
56 Qwen3.7-Plus 15.8% Pareto chart; effort unspecified Jun 2026 Official leaderboard
57 Muse Glimmer 15.4% xhigh Aug 2026 Official leaderboard
58 Inkling 15% high Jul 2026 Official leaderboard
59 Grok 4.3 12.9% Pareto chart; high Apr 2026 Official leaderboard
60 Kimi K2.5 12.6% Pareto chart; effort unspecified Jan 2026 Official leaderboard
61 Kimi K2.6 12.2% Pareto chart; effort unspecified Apr 2026 Official leaderboard
62 Gemini 3.5 Flash-Lite 10.2% Pareto chart; effort unspecified Jul 2026 Official leaderboard
63 Mistral Large 3 9.00% Pareto chart; effort unspecified Dec 2025 Official leaderboard
64 Nemotron 3 Ultra 6.80% Jun 2026 Official leaderboard

Labs report scores for 18 models in their cards. Switch the scores above to rank them together.

Score vs. generated tokens

The published Pareto chart snapshot is separate from the current leaderboard. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 10% 20% 30% 40% 50% 1001,00010K100K Generated tokens per trial, log scale, lower is better Score (pass@1) GPT-5.6 Sol (Pareto chart; max)GPT-5.6 Sol (Pareto chart; effort unspecified)Claude Fable 5 (Pareto chart; effort unspecified)GPT-5.4 (Pareto chart; effort unspecified) 0% 10% 20% 30% 40% 50% 1001,00010K100K Generated tokens per trial, log scale, lower is better Score (pass@1) GPT-5.6 Sol (Pareto chart; max)GPT-5.6 Sol (Pareto chart; effort unspecified)Claude Fable 5 (Pareto chart; effort unspecified)GPT-5.4 (Pareto chart; effort unspecified)
ModelSettingScore (pass@1)Generated tokens per trial
GPT-5.6 Sol Pareto chart; max 45% 9,645
GPT-5.6 Sol Pareto chart; effort unspecified 39.5% 1,764
Gemini 3.5 Flash Pareto chart; effort unspecified 35.9% 5,617
Claude Fable 5 Pareto chart; max 34.8% 6,590
Gemini 3.6 Flash Pareto chart; effort unspecified 34% 4,648
GPT-5.6 Terra Pareto chart; max 34% 15,324
GPT-5.6 Luna Pareto chart; max 31.4% 20,713
GPT-5.5 Pareto chart; xhigh 31% 9,006
GPT-5.4 Pareto chart; xhigh 29.6% 9,464
Claude Fable 5 Pareto chart; effort unspecified 29.5% 1,293
GPT-5.5 Pareto chart; effort unspecified 28.5% 2,661
GPT-5.6 Terra Pareto chart; effort unspecified 26.6% 1,293
Kimi K3 Pareto chart; effort unspecified 26.6% 18,393
Gemini 3.1 Pro Pareto chart; effort unspecified 26.1% 8,175
Muse Spark 1.1 Pareto chart; xhigh 24.4% 8,086
Muse Spark 1.1 Pareto chart; effort unspecified 23.6% 8,298
GPT-5.6 Luna Pareto chart; effort unspecified 21.4% 1,832
Grok 4.5 Pareto chart; effort unspecified 17.3% 13,228
Grok 4.5 Pareto chart; high 16.7% 13,441
Claude Sonnet 5 Pareto chart; max 16.6% 32,258
Claude Opus 4.7 Pareto chart; max 16.5% 12,786
Claude Opus 4.8 Pareto chart; max 15.9% 6,031
Qwen3.5-Plus Pareto chart; effort unspecified 15.9% 16,663
Qwen3.7-Plus Pareto chart; effort unspecified 15.8% 10,676
GPT-5.4 Pareto chart; effort unspecified 14.1% 130.38
Claude Opus 4.7 Pareto chart; effort unspecified 13.6% 545.56
Grok 4.3 Pareto chart; high 12.9% 8,997
Kimi K2.5 Pareto chart; effort unspecified 12.6% 15,622
Claude Sonnet 5 Pareto chart; effort unspecified 12.2% 2,107
Kimi K2.6 Pareto chart; effort unspecified 12.2% 28,798
Grok 4.3 Pareto chart; effort unspecified 11.7% 1,432
Claude Opus 4.8 Pareto chart; effort unspecified 11.3% 363.78
Gemini 3.5 Flash-Lite Pareto chart; effort unspecified 10.2% 285.88
Mistral Large 3 Pareto chart; effort unspecified 9% 284.68
Inkling Pareto chart; effort unspecified 3.4% 2,953

Of runs with a cost, the best scores 45% for $0.29 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Mistral AI OpenAI xAI Google DeepMind Cost frontier One model at each effort level
0% 10% 20% 30% 40% 50% $0.0001 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale GPT-5.6 Sol, Pareto chart; max GPT-5.6 Sol, Pareto chart; effort unspecified Gemini 3.5 Flash, Pareto chart; effort unspecified GPT-5.6 Terra, Pareto chart; effort unspecified GPT-5.6 Luna, Pareto chart; effort unspecified GPT-5.4, Pareto chart; effort unspecified Grok 4.3, Pareto chart; effort unspecified 0% 10% 20% 30% 40% 50% $0.0001 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale GPT-5.6 Sol, Pareto chart; max GPT-5.6 Sol, Pareto chart; effort unspecified

Source: Official Chartography leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-6.2 Astra max 90.7%
Claude Opus 5.5 adaptive/max 89.6%
GPT-6 Astra max 88.6%
GPT-6.2 Sol max 87.7%
Gemini 4 Hydrogen xhigh 87.2%
Gemini 4 Neon high 86.6%
Gemini 4 Carbon high 85.5%
Muse Spark 1.3 max 84.6%
Claude Opus 5 adaptive/max 84.4%
Gemini 3.8 Flash high 81.7%
Claude Fable 5.2 adaptive/max 80.8%
Claude Fable 5.1 adaptive/max 79.6%
GPT-6 Sol max 79.4%
Muse Ember max 79.3%
Claude Fable 5 adaptive/max 79%
GPT-5.6 Sol max 78.6%
DeepSeek-V4-Flash max 77.1%
Inkling 2 Small xhigh 75.6%
Qwen4-Max xhigh 75.3%
Qwen 3.8 Max xhigh 75.2%
Gemini 3.7 Flash high 74.9%
Grok 4.7 xhigh 74.6%
GLM-5.3-Flash max 72.7%
GPT-5.6 Luna max 72.2%
GPT-6 Luna max 68%
Muse Spark 1.2 xhigh 66.7%
Kimi K3 max 66.4%
GLM-5.3 max 64.2%
Claude Sonnet 5 adaptive/max 64%
Tencent Hy4 preview high 59.4%
GPT-5.6 Terra xhigh 58.8%
Gemini 3.1 Pro high 54.2%
Tencent Hy4 high 49.2%
GPT-5.6 Sol Pareto chart; max 45% $0.29 9,645
Grok 4.6 xhigh 44.4%
MiniMax-M3.1-Flash max 39.8%
GPT-5.6 Sol Pareto chart; effort unspecified 39.5% $0.06 1,764
Gemini 3.5 Flash Pareto chart; effort unspecified 35.9% $0.05 5,617
Claude Fable 5 Pareto chart; max 34.8% $0.35 6,590
GPT-5.6 Terra Pareto chart; max 34% $0.23 15,324
Gemini 3.6 Flash Pareto chart; effort unspecified 34% $0.04 4,648
Kimi K2.7 Code max 31.5%
GPT-5.6 Luna Pareto chart; max 31.4% $0.13 20,713
GPT-5.5 Pareto chart; xhigh 31% $0.28 9,006
GPT-5.4 Pareto chart; xhigh 29.6% $0.14 9,464
Claude Fable 5 Pareto chart; effort unspecified 29.5% $0.09 1,293
GPT-5.5 Pareto chart; effort unspecified 28.5% $0.09 2,661
GPT-5.6 Terra Pareto chart; effort unspecified 26.6% $0.02 1,293
Kimi K3 Pareto chart; effort unspecified 26.6% $0.28 18,393
Gemini 3.1 Pro Pareto chart; effort unspecified 26.1% $0.10 8,174
Muse Spark 1.1 Pareto chart; xhigh 24.4% $0.05 8,086
Muse Spark 1.1 Pareto chart; effort unspecified 23.6% $0.04 8,298
GPT-5.6 Luna Pareto chart; effort unspecified 21.4% $0.01 1,832
DeepSeek-V4-Pro max 21.1%
Tencent Hy3 high 20.3%
Grok 4.5 Pareto chart; effort unspecified 17.3% $0.08 13,228
Grok 4.5 Pareto chart; high 16.7% $0.08 13,441
Claude Sonnet 5 Pareto chart; max 16.6% $0.49 32,258
Claude Opus 4.7 Pareto chart; max 16.5% $0.33 12,786
Qwen3.5-Plus Pareto chart; effort unspecified 15.9% $0.04 16,663
Claude Opus 4.8 Pareto chart; max 15.9% $0.16 6,031
Qwen3.7-Plus Pareto chart; effort unspecified 15.8% $0.01 10,676
Muse Glimmer xhigh 15.4%
Inkling high 15%
GPT-5.4 Pareto chart; effort unspecified 14.1% $0.004 130
Claude Opus 4.7 Pareto chart; effort unspecified 13.6% $0.02 546
Grok 4.3 Pareto chart; high 12.9% $0.02 8,997
Kimi K2.5 Pareto chart; effort unspecified 12.6% $0.03 15,622
Kimi K2.6 Pareto chart; effort unspecified 12.2% $0.10 28,798
Claude Sonnet 5 Pareto chart; effort unspecified 12.2% $0.04 2,107
Grok 4.3 Pareto chart; effort unspecified 11.7% $0.004 1,432
Claude Opus 4.8 Pareto chart; effort unspecified 11.3% $0.02 364
Gemini 3.5 Flash-Lite Pareto chart; effort unspecified 10.2% $0.001 286
Mistral Large 3 Pareto chart; effort unspecified 9.00% $0.0008 285
Mistral Large 3 7.30%
Nemotron 3 Ultra 6.80%
Inkling Pareto chart; effort unspecified 3.40% $0.01 2,953

5 models have held number one on the official leaderboard since its release in August 2026 6 models have held number one with claims included since its release in August 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 and what labs claim in their cards.

Anthropic Meta OpenAI DeepSeek Z.ai Moonshot AI
Own leaderboard Passed 85% in Sep 2026, by GPT-6 Astra
Lab claims Passed 85% in Sep 2026, by Claude Opus 5.5

Sources: Official Chartography leaderboard and labs’ model cards

GPT-6 Astra on the official leaderboard passed 85% in September 2026

That was 1 month after its release. By the same age, 20% of the benchmarks from 2026 had passed 85%. 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 Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra Muse Spark 1.3 Claude Fable 5 Gemini 3.1 Pro Qwen3.5-Plus Mistral Large 3 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) GPT-6.2 Astra Claude Opus 5

Sources: Official Chartography leaderboard and labs’ model cards

7 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.

DeepSeek
Z.ai
Anthropic
Google DeepMind
Moonshot AI
Thinking Machines
Qwen
DateLaunch postScoreAs the post writes it
Nov 20, 2026 Inkling 2 Mini 22.1 Chartography (original, 72.5 with Python)
Nov 18, 2026 Gemini 4 Neon 73.5 Chartography
Nov 15, 2026 Qwen4-Max 54.2 Visual chart understanding, Chartography (min_pixels 4096×32×32)
Nov 12, 2026 Gemini 4 Hydrogen 66.6 Chartography
Nov 9, 2026 Claude Fable 5.2 73.4 Visual chart recognition Chartography (no tools), max effort. With tools 91.6% (launch table and Table 8.1.A)
Nov 5, 2026 Inkling 2 Small 36.7 Chartography (original; 80.4 with Python)
Oct 28, 2026 Gemini 4 Carbon 72.5 Chartography
Oct 24, 2026 Kimi K3.1 Pro 90.2 ChartoGraphy w/ python
Oct 17, 2026 DeepSeek V4.1 Flash 1017 83 Chartography (w/tools)
Oct 13, 2026 Gemini 4 Argon 71.6 Chartography
Oct 7, 2026 Claude Haiku 5.5 46.4 Visual reasoning Chartography (no tools)
Sep 28, 2026 Claude Sonnet 5.5 61.6 Visual chart recognition Chartography (no tools)
Sep 22, 2026 Claude Opus 5.5 89 Visual chart recognition Chartography (with tools)
Sep 10, 2026 DeepSeek V4.1 Flash 78.9 Chartography w/ tools (Pass@1)
Aug 26, 2026 GLM-5.3-Flash 78 Chartography (w/ Tools)
Aug 21, 2026 DeepSeek-V4-Flash-Vision-Exp 64.3 Chartography

11 system cards report it, from 4 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
Nov 20, 2026 Inkling 2 Mini model card 22.1 Chartography (original, 72.5 with Python) Capabilities
Nov 18, 2026 Gemini 4 Neon model card 73.5 Chartography (results table, p.4) Capabilities
Nov 10, 2026 Gemini 4 Carbon model card 72.5 Chartography (results table) Capabilities
Nov 9, 2026 Claude Fable 5.2 system card 73.4 Chartography, no tools, adaptive thinking max effort, five runs, Gemini 3.5 Flash grader. With tools 91.6%. Separate contamination check: original charts 74.3%, redrawn charts 76.4% (Table 8.1.A and Section 8.13.1) Capabilities
Oct 13, 2026 Gemini 4 Argon model card 71.6 Chartography (results table, p.5) Capabilities
Oct 7, 2026 Claude Haiku 5.5 system card 46.4 Chartography, no tools, max effort (with tools 86.2) (8.9.1) Capabilities
Sep 28, 2026 Claude Sonnet 5.5 system card 61.6 Chartography without tools, max effort (90.2% with tools) (8.13.1, Figure 8.13.1.A) Capabilities
Sep 22, 2026 Claude Opus 5.5 system card 64.4 Chartography, no tools, adaptive thinking at max effort, five runs (89.0% with tools), Gemini 3.5 Flash grader (8.13.1, Figure 8.13.1.A) Capabilities
Sep 10, 2026 DeepSeek V4.1 Flash technical report 78.9 Chartography w/ tools (Pass@1) (Table 3) Capabilities
Sep 1, 2026 Claude Fable 5.1 system card 42.6 Chartography, no tools, adaptive thinking max effort, 5 runs; 86.2% with tools (8.14.1) Capabilities
Jul 24, 2026 Claude Opus 5 system card 29.6 Chartography, no tools, adaptive thinking at max effort, five runs (83.0% with tools) (8.12.1) Capabilities

The lab's number comes out ahead in 1 of 10 pairs that differ

Each score a lab reported for its own model, beside the same model's best score on the leaderboard. Gaps mostly come from different harnesses, effort levels and task subsets. Every benchmark.

Model Lab says Board says Gap What differs
Claude Opus 5 29.6% Chartography, no tools, adaptive thinking at max effort, five runs (83.0% with tools) (8.12.1) 84.4% Leaderboard, adaptive/max −55 pts
Inkling 2 Small 36.7% Chartography (original; 80.4 with Python) 75.6% Leaderboard, xhigh −39 pts
Claude Fable 5.1 42.6% Chartography, no tools, adaptive thinking max effort, 5 runs; 86.2% with tools (8.14.1) 79.6% Leaderboard, adaptive/max −37 pts
Qwen4-Max 54.2% Visual chart understanding, Chartography (min_pixels 4096×32×32) 75.3% Leaderboard, xhigh −21 pts
Gemini 4 Hydrogen 66.6% Chartography 87.2% Leaderboard, xhigh −21 pts
Gemini 4 Neon 73.5% Chartography 86.6% Leaderboard, high −13 pts
Gemini 4 Carbon 72.5% Chartography 85.5% Leaderboard, high −13 pts
Claude Fable 5.2 73.4% Visual chart recognition Chartography (no tools), max effort. With tools 91.6% (launch table and Table 8.1.A) 80.8% Leaderboard, adaptive/max −7.4 pts
GLM-5.3-Flash 78% Chartography (w/ Tools) 72.7% Leaderboard, max +5.3 pts
Claude Opus 5.5 89% Visual chart recognition Chartography (with tools) 89.6% Leaderboard, adaptive/max −0.6 pts Anthropic reports Chartography with tools at max effort. The compared board configuration evaluates chart responses without tools.

More in Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%

Data audit

9 of 11 checks pass, 2 warnings. Look first at the Lab claims check. 7 of 10 pairs differ by more than 10 points, the largest for Claude Opus 5 (−55 pts). Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (56 entries), launch posts (16 scores) and system cards (11 scores).
  • Pass Models scored 64 models we track.
  • Pass Newest score The newest is from November 2026, for Inkling 2 Mini.
  • Pass Frontier models 8 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Note Headroom Saturated. The best is 90.7%, past 85% since September 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on surgehq.ai: 77 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 warning 1 pass
  • Warning Lab claims 7 of 10 pairs differ by more than 10 points, the largest for Claude Opus 5 (−55 pts).
  • Pass Known issues No published audit found problems.

How we check