Benchmark Almanac

CAIS Text Capabilities Index

Gemini 4 Argon 66.5 leads since September 2026
Not in model cards

Averages model scores on the text tests that the Center for AI Safety runs for its AI Dashboard. On November 1st, 2026 the index listed Humanity's Last Exam and TextQuests.

One of three dashboard indexes, next to the Vision Capabilities Index and the Risk Index. The dashboard data also hold ARC-AGI-2, SWE-Bench Pro and HLE-Diamond scores. 67 models had entries on November 1st, 2026.

Released
December 2025
Kind
Composite indices
Level
Research frontier
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Strongest model tested
Claude Fable 5.2 #1 on the capability index
Last model added
November 2026
Scores refreshed
November 12th, 2026
68
Models scored on its own leaderboard

Gemini 4 Argon leads with 66.5

Best score per model from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best.

68 models. Scroll sideways for the rest.

  1. 66.5 Gemini 4 Argon
  2. 66.3 Claude Fable 5.2
  3. 63.9 GPT-6 Astra
  4. 61.4 GPT-6.1 Sol
  5. 59.0 Gemini 4 Carbon
  6. 56.3 Claude Opus 5.5
  7. 55.6 Claude Fable 5.1
  8. 54.4 Claude Fable 5
  9. 52.2 Claude Opus 5
  10. 50.2 Gemini 3.8 Flash
  11. 49.3 Gemini 3.7 Flash
  12. 48.8 GPT-5.6 Sol
  13. 47.6 Muse Wonder
  14. 46.4 Kimi K3.1 Pro
  15. 46.3 GPT-5.5
  16. 45.8 Gemini 3.1 Pro
  17. 43.5 Grok 4.6
  18. 41.3 GPT-5.4
  19. 41.2 Claude Opus 4.8
  20. 40.5 Kimi K3
  21. 39.7 Gemini 3 Pro
  22. 39.1 Gemini 3.5 Flash
  23. 38.2 GPT-5.6 Terra
  24. 38.1 Muse Spark 1.3
  25. 38.0 Claude Opus 4.7
  26. 37.5 Claude Opus 4.6
  27. 37.4 GLM-5.2
  28. 37.2 Muse Spark 1.1
  29. 36.5 Gemini 3 Flash
  30. 36.3 Muse Spark Open
  31. 35.6 GLM-5.3
  32. 33.5 GPT-5.2
  33. 32.3 Claude Opus 4.5
  34. 32.3 Grok 4.5
  35. 31.4 Kimi K2.6
  36. 31.3 GPT-5.6 Luna
  37. 30.7 GPT-5.1
  38. 27.7 GPT-5
  39. 27.6 GLM-5.1
  40. 26.6 GPT-5.4 Mini
  41. 26.4 DeepSeek-V4-Pro
  42. 26.3 Claude Sonnet 4.6
  43. 26.2 Grok 4
  44. 25.6 o3
  45. 24.9 Kimi K2.5
  46. 24.3 Grok 4.20
  47. 23.4 Grok 4.3
  48. 22.9 Claude Sonnet 4.5
  49. 22.4 Gemini 2.5 Pro
  50. 21.5 DeepSeek-V3.2
  51. 19.9 Kimi K2 Thinking
  52. 19.0 Grok 4 Fast
  53. 18.8 Gemini 2.5 Pro
  54. 18.5 GPT-5 mini
  55. 18.1 Gemini 3.1 Flash-Lite
  56. 17.4 Grok 4.1 Fast
  57. 17.1 Claude Sonnet 4
  58. 16.1 GPT-5.4 Nano
  59. 15.4 o4-mini
  60. 13.2 Gemini 2.5 Flash
  61. 12.7 o3-mini
  62. 12.4 Claude Haiku 4.5
  63. 11.8 DeepSeek-R1
  64. 9.20 Gemini 2.5 Flash-Lite
  65. 8.00 o1
  66. 8.00 Claude 3.7 Sonnet
  67. 7.90 GPT-4o
  68. 6.60 GPT-5 nano
  1. 1 Gemini 4 Argon high 66.5
  2. 2 Claude Fable 5.2 high 66.3
  3. 3 GPT-6 Astra high 63.9
  4. 4 GPT-6.1 Sol high 61.4
  5. 5 Gemini 4 Carbon high 59.0
  6. 6 Claude Opus 5.5 high 56.3
  7. 7 Claude Fable 5.1 high 55.6
  8. 8 Claude Fable 5 high 54.4
  9. 9 Claude Opus 5 high 52.2
  10. 10 Gemini 3.8 Flash high 50.2
  11. 11 Gemini 3.7 Flash high 49.3
  12. 12 GPT-5.6 Sol high 48.8
  13. 13 Muse Wonder high 47.6
  14. 14 Kimi K3.1 Pro high 46.4
  15. 15 GPT-5.5 high 46.3
  16. 16 Gemini 3.1 Pro high 45.8
  17. 17 Grok 4.6 43.5
  18. 18 GPT-5.4 high 41.3
  19. 19 Claude Opus 4.8 high 41.2
  20. 20 Kimi K3 high 40.5
  21. 21 Gemini 3 Pro high 39.7
  22. 22 Gemini 3.5 Flash 39.1
  23. 23 GPT-5.6 Terra high 38.2
  24. 24 Muse Spark 1.3 high 38.1
  25. 25 Claude Opus 4.7 38.0
  26. 26 Claude Opus 4.6 37.5
  27. 27 GLM-5.2 37.4
  28. 28 Muse Spark 1.1 high 37.2
  29. 29 Gemini 3 Flash high 36.5
  30. 30 Muse Spark Open high 36.3
  31. 31 GLM-5.3 35.6
  32. 32 GPT-5.2 high 33.5
  33. 33 Claude Opus 4.5 thinking 32.3
  34. 34 Grok 4.5 32.3
  35. 35 Kimi K2.6 31.4
  36. 36 GPT-5.6 Luna high 31.3
  37. 37 GPT-5.1 high 30.7
  38. 38 GPT-5 high 27.7
  39. 39 GLM-5.1 27.6
  40. 40 GPT-5.4 Mini high 26.6
  41. 41 DeepSeek-V4-Pro 26.4
  42. 42 Claude Sonnet 4.6 26.3
  43. 43 Grok 4 26.2
  44. 44 o3 high 25.6
  45. 45 Kimi K2.5 thinking 24.9
  46. 46 Grok 4.20 24.3
  47. 47 Grok 4.3 23.4
  48. 48 Claude Sonnet 4.5 thinking 22.9
  49. 49 Gemini 2.5 Pro Jun 2025, high 22.4
  50. 50 DeepSeek-V3.2 thinking 21.5
  51. 51 Kimi K2 Thinking thinking 19.9
  52. 52 Grok 4 Fast 19.0
  53. 53 Gemini 2.5 Pro Mar 2025, high 18.8
  54. 54 GPT-5 mini high 18.5
  55. 55 Gemini 3.1 Flash-Lite high 18.1
  56. 56 Grok 4.1 Fast 17.4
  57. 57 Claude Sonnet 4 thinking 17.1
  58. 58 GPT-5.4 Nano high 16.1
  59. 59 o4-mini high 15.4
  60. 60 Gemini 2.5 Flash Jun 2025, high 13.2
  61. 61 o3-mini high 12.7
  62. 62 Claude Haiku 4.5 thinking 12.4
  63. 63 DeepSeek-R1 11.8
  64. 64 Gemini 2.5 Flash-Lite Jun 2025, high 9.20
  65. 65 o1 8.00
  66. 66 Claude 3.7 Sonnet thinking 8.00
  67. 67 GPT-4o Nov 2024 7.90
  68. 68 GPT-5 nano high 6.60

Source: Official CAIS Text Capabilities Index leaderboard

#ModelScoreSettingDateSource
1 Gemini 4 Argon 66.5 high Sep 2026 Official leaderboard
2 Claude Fable 5.2 66.3 high Nov 2026 Official leaderboard
3 GPT-6 Astra 63.9 high Sep 2026 Official leaderboard
4 GPT-6.1 Sol 61.4 high Sep 2026 Official leaderboard
5 Gemini 4 Carbon 59.0 high Oct 2026 Official leaderboard
6 Claude Opus 5.5 56.3 high Sep 2026 Official leaderboard
7 Claude Fable 5.1 55.6 high Sep 2026 Official leaderboard
8 Claude Fable 5 54.4 high Jun 2026 Official leaderboard
9 Claude Opus 5 52.2 high Jul 2026 Official leaderboard
10 Gemini 3.8 Flash 50.2 high Sep 2026 Official leaderboard
11 Gemini 3.7 Flash 49.3 high Aug 2026 Official leaderboard
12 GPT-5.6 Sol 48.8 high Jul 2026 Official leaderboard
13 Muse Wonder 47.6 high Oct 2026 Official leaderboard
14 Kimi K3.1 Pro 46.4 high Oct 2026 Official leaderboard
15 GPT-5.5 46.3 high Apr 2026 Official leaderboard
16 Gemini 3.1 Pro 45.8 high Feb 2026 Official leaderboard
17 Grok 4.6 43.5 Aug 2026 Official leaderboard
18 GPT-5.4 41.3 high Mar 2026 Official leaderboard
19 Claude Opus 4.8 41.2 high May 2026 Official leaderboard
20 Kimi K3 40.5 high Jul 2026 Official leaderboard
21 Gemini 3 Pro 39.7 high Nov 2025 Official leaderboard
22 Gemini 3.5 Flash 39.1 May 2026 Official leaderboard
23 GPT-5.6 Terra 38.2 high Jul 2026 Official leaderboard
24 Muse Spark 1.3 38.1 high Sep 2026 Official leaderboard
25 Claude Opus 4.7 38.0 Apr 2026 Official leaderboard
26 Claude Opus 4.6 37.5 Feb 2026 Official leaderboard
27 GLM-5.2 37.4 Jun 2026 Official leaderboard
28 Muse Spark 1.1 37.2 high Jul 2026 Official leaderboard
29 Gemini 3 Flash 36.5 high Dec 2025 Official leaderboard
30 Muse Spark Open 36.3 high Oct 2026 Official leaderboard
31 GLM-5.3 35.6 Aug 2026 Official leaderboard
32 GPT-5.2 33.5 high Dec 2025 Official leaderboard
33 Claude Opus 4.5 32.3 thinking Nov 2025 Official leaderboard
34 Grok 4.5 32.3 Jul 2026 Official leaderboard
35 Kimi K2.6 31.4 Apr 2026 Official leaderboard
36 GPT-5.6 Luna 31.3 high Jul 2026 Official leaderboard
37 GPT-5.1 30.7 high Nov 2025 Official leaderboard
38 GPT-5 27.7 high Aug 2025 Official leaderboard
39 GLM-5.1 27.6 Apr 2026 Official leaderboard
40 GPT-5.4 Mini 26.6 high Mar 2026 Official leaderboard
41 DeepSeek-V4-Pro 26.4 Apr 2026 Official leaderboard
42 Claude Sonnet 4.6 26.3 Feb 2026 Official leaderboard
43 Grok 4 26.2 Jul 2025 Official leaderboard
44 o3 25.6 high Apr 2025 Official leaderboard
45 Kimi K2.5 24.9 thinking Jan 2026 Official leaderboard
46 Grok 4.20 24.3 Feb 2026 Official leaderboard
47 Grok 4.3 23.4 Apr 2026 Official leaderboard
48 Claude Sonnet 4.5 22.9 thinking Sep 2025 Official leaderboard
49 Gemini 2.5 Pro Jun 2025 22.4 high Jun 2025 Official leaderboard
50 DeepSeek-V3.2 21.5 thinking Dec 2025 Official leaderboard
51 Kimi K2 Thinking 19.9 thinking Nov 2025 Official leaderboard
52 Grok 4 Fast 19.0 Sep 2025 Official leaderboard
53 Gemini 2.5 Pro Mar 2025 18.8 high Mar 2025 Official leaderboard
54 GPT-5 mini 18.5 high Aug 2025 Official leaderboard
55 Gemini 3.1 Flash-Lite 18.1 high Mar 2026 Official leaderboard
56 Grok 4.1 Fast 17.4 Nov 2025 Official leaderboard
57 Claude Sonnet 4 17.1 thinking May 2025 Official leaderboard
58 GPT-5.4 Nano 16.1 high Mar 2026 Official leaderboard
59 o4-mini 15.4 high Apr 2025 Official leaderboard
60 Gemini 2.5 Flash Jun 2025 13.2 high Jun 2025 Official leaderboard
61 o3-mini 12.7 high Jan 2025 Official leaderboard
62 Claude Haiku 4.5 12.4 thinking Oct 2025 Official leaderboard
63 DeepSeek-R1 11.8 Jan 2025 Official leaderboard
64 Gemini 2.5 Flash-Lite Jun 2025 9.20 high Jun 2025 Official leaderboard
65 o1 8.00 Dec 2024 Official leaderboard
66 Claude 3.7 Sonnet 8.00 thinking Feb 2025 Official leaderboard
67 GPT-4o Nov 2024 7.90 Nov 2024 Official leaderboard
68 GPT-5 nano 6.60 high Aug 2025 Official leaderboard

7 models have held number one on the official leaderboard since its release in December 2025

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

Google DeepMind OpenAI Anthropic
Own leaderboard 7 leaders
Gemini 4 Argon 66.5

Source: Official CAIS Text Capabilities Index leaderboard

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
0.00 20.0 40.0 60.0 80.0 2025 2026 2027 0.00 20.0 40.0 60.0 80.0 2025 2026 2027

Source: Official CAIS Text Capabilities Index leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Other

All 43 in Other
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Gemini 4 Carbon 59.0
  • SAGE v1.1 (Vals) In use Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading. Claude Opus 5.5 91.3%
  • MTEB Fading Massive Text Embedding Benchmark: retrieval, clustering, classification across languages and modalities. sonar 42%
  • Anthropic ECI (AECI) Only in system cards Index that combines a basket of capability benchmarks into one capability score to track trends across Anthropic models. Claude Opus 5.5 169.4
  • SPEED-Bench Only in system cards Measures speculative decoding across varied prompts and domains. Reports the average acceptance length of draft tokens, so higher means faster generation at the same output. Nemotron 3 Ultra 5.15