Benchmark Almanac

AGI-Eval Community Image and Video Understanding

Seed-2.0-Pro-260215 86.7 tops its leaderboard Gemini 3 Flash 86.5 leads the models we track
Not in model cards

Models interpret images and videos. The board reports visual question accuracy and separate comparison win rates.

The 2026/11 leaderboard reports 96 scored model entries on its primary measure.

Released
February 2025
Kind
General visual questions
Level
Everyday
Scored
Checked automatically
Answers
Short answer
Tasks
Kept private
Measures
Capability
In and out
Images, Video, Text → Text
Interaction
Single turn
Refresh
One fixed set
Run by
A host
In views
Vision
Strongest model tested
Claude Opus 4.6 #39 on the capability index
Last model added
April 2026
Scores refreshed
November 21st, 2026
91
Entries on its own leaderboard, 63 models we track

Seed-2.0-Pro-260215 tops the leaderboard with 86.7 Gemini 3 Flash leads with 86.5 among models released by April 2026

Best score per entry from the benchmark's own leaderboard, read on November 21st, 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. The best model we track is Gemini 3 Flash, at 86.5. This view shows only models we track, each at its best.

The newest model with a score here came out in April 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

91 entries. Scroll sideways for the rest.

  1. 86.7 Seed-2.0-Pro-260215
  2. 86.5 Gemini 3 Flash
  3. 86.2 Gemini-3-Pro-Preview-Multimodal
  4. 86.2 Qwen3.5-Plus
  5. 85.9 Gemini 3.1 Pro
  6. 84.3 Kimi K2.6
  7. 82.1 Seed1.8
  8. 81.8 Qwen3.5-Flash
  9. 81.7 Kimi K2.5
  10. 81.1 Seed1.5-VL
  11. 80.1 Gemini 2.5 Pro
  12. 79.8 Qwen3-VL-235B-A22B
  13. 78.3 Gemini 3.1 Flash-Lite
  14. 78.2 GPT-5
  15. 77.7 Doubao-Seed-1.6-Thinking-2507
  16. 77.3 Claude Opus 4.6
  17. 76.9 GLM-4.6V
  18. 76.8 Qwen3-VL-235B-A22B-Instruct
  19. 76.6 Gemini 2.5 Flash
  20. 76.5 o3
  21. 76.3 Doubao-Seed-1.6-Thinking-2506
  22. 75.5 Doubao-Seed-1.6-2506
  23. 75.1 Claude Sonnet 4.6
  24. 74.8 Doubao-1.5-Vision-Pro
  25. 74.5 Qwen3-VL-32B-Instruct
  26. 73.7 o4-mini
  27. 73.6 Doubao-Seed-1.6-Flash-250615
  28. 73.4 Qwen3-VL-32B-Thinking
  29. 73.3 Doubao-1.5-Vision-Pro-32k-250115
  30. 71.9 GLM-4.5V
  31. 71.6 MiMo-VL-7B-RL-2508
  32. 71.5 Intern-S1
  33. 71.4 Qwen3-Omni-Flash-2025-1201
  34. 71.2 Step 3
  35. 71.0 Kimi-Thinking-Preview
  36. 70.5 Qwen3-Omni-30B-A3B
  37. 69.9 Kimi-VL-A3B-Thinking-2506
  38. 69.6 MiMo-VL-7B-SFT-2508
  39. 69.5 GLM-4.6V-Flash
  40. 68.9 GPT-4.1
  41. 68.8 Gemini 2.0 Pro
  42. 68.3 GLM-4.1V-9B-Thinking
  43. 68.0 Doubao-1.5-Vision-Lite-250315
  44. 67.9 GPT-4.5
  45. 67.5 Claude Opus 4
  46. 67.3 Llama 4 Maverick
  47. 67.3 Qwen3-VL-8B-Thinking
  48. 67.0 InternVL3.5-241B-A28B
  49. 67.0 Qwen3-VL-8B-Instruct
  50. 67.0 MiMo-VL-7B-RL
  51. 66.8 Gemini 2.0 Flash
  52. 66.3 MiMo-VL-7B-SFT
  53. 65.2 InternVL3.5-38B
  54. 65.0 Claude Sonnet 4
  55. 64.8 InternVL3-78B
  56. 64.5 SenseChat-SenseNova-V6-Pro
  57. 64.3 Skywork-R1V3-38B
  58. 64.2 GLM-4V
  59. 63.4 Step-1o-vision-32k
  60. 62.6 GPT-4o-1120
  61. 62.5 Qwen2.5-VL-32B
  62. 62.4 SenseChat-Vision-V5.5
  63. 61.6 Keye-VL-1.5-8B
  64. 61.6 Llama 4 Scout
  65. 61.5 Qwen2.5-72B-VL-Instruct
  66. 61.4 Doubao-Pro-32k
  67. 61.4 InternVL3.5-30B-A3B
  68. 61.1 MiniCPM-V 4.5
  69. 60.8 Claude 3.7 Sonnet
  70. 60.5 InternVL3.5-14B
  71. 59.5 GPT-4o-0513
  72. 59.2 InternVL3.5-8B
  73. 59.1 Claude-3.5-Sonnet-1022
  74. 58.2 Skywork-R1V-38B
  75. 58.0 InternVL2_5-78B
  76. 56.8 Keye-VL-8B-Preview-Auto
  77. 55.9 Keye-VL-8B-Preview
  78. 55.9 BLIP3o-8B
  79. 55.0 Gemma 3 27B
  80. 54.5 InternVL3.5-GPT-OSS-20B-A4B-Preview
  81. 54.0 Hunyuan-Vision-20250103
  82. 53.9 BAGEL-7B-MoT
  83. 53.4 Qwen2.5-VL-7B
  84. 52.9 Qwen-Omni-Turbo
  85. 51.1 Gemma 3 12B
  86. 49.7 Ming-Lite-Omni
  87. 48.7 Ming-Lite-Omni-1.5
  88. 45.7 Qwen2.5-Omni-3B
  89. 42.4 Qwen2.5-Omni-7B
  90. 41.9 MiniCPM-o-2_6
  91. 40.0 Gemma 3 4B
  1. 1 Seed-2.0-Pro-260215 86.7
  2. 2 Gemini 3 Flash 86.5
  3. 3 Gemini-3-Pro-Preview-Multimodal 86.2
  4. 4 Qwen3.5-Plus 86.2
  5. 5 Gemini 3.1 Pro 85.9
  6. 6 Kimi K2.6 84.3
  7. 7 Seed1.8 82.1
  8. 8 Qwen3.5-Flash 81.8
  9. 9 Kimi K2.5 81.7
  10. 10 Seed1.5-VL 81.1
  11. 11 Gemini 2.5 Pro Jun 2025 80.1
  12. 12 Qwen3-VL-235B-A22B thinking 79.8
  13. 13 Gemini 3.1 Flash-Lite 78.3
  14. 14 GPT-5 78.2
  15. 15 Doubao-Seed-1.6-Thinking-2507 77.7
  16. 16 Claude Opus 4.6 77.3
  17. 17 GLM-4.6V 76.9
  18. 18 Qwen3-VL-235B-A22B-Instruct 76.8
  19. 19 Gemini 2.5 Flash Jun 2025, non-thinking 76.6
  20. 20 o3 76.5
  21. 21 Doubao-Seed-1.6-Thinking-2506 76.3
  22. 22 Doubao-Seed-1.6-2506 75.5
  23. 23 Claude Sonnet 4.6 75.1
  24. 24 Doubao-1.5-Vision-Pro 74.8
  25. 25 Qwen3-VL-32B-Instruct 74.5
  26. 26 o4-mini 73.7
  27. 27 Doubao-Seed-1.6-Flash-250615 73.6
  28. 28 Qwen3-VL-32B-Thinking 73.4
  29. 29 Doubao-1.5-Vision-Pro-32k-250115 73.3
  30. 30 GLM-4.5V 71.9
  31. 31 MiMo-VL-7B-RL-2508 nothinking 71.6
  32. 32 Intern-S1 71.5
  33. 33 Qwen3-Omni-Flash-2025-1201 71.4
  34. 34 Step 3 71.2
  35. 35 Kimi-Thinking-Preview 71.0
  36. 36 Qwen3-Omni-30B-A3B 70.5
  37. 37 Kimi-VL-A3B-Thinking-2506 69.9
  38. 38 MiMo-VL-7B-SFT-2508 69.6
  39. 39 GLM-4.6V-Flash 69.5
  40. 40 GPT-4.1 68.9
  41. 41 Gemini 2.0 Pro 68.8
  42. 42 GLM-4.1V-9B-Thinking 68.3
  43. 43 Doubao-1.5-Vision-Lite-250315 68.0
  44. 44 GPT-4.5 67.9
  45. 45 Claude Opus 4 67.5
  46. 46 Llama 4 Maverick 67.3
  47. 47 Qwen3-VL-8B-Thinking 67.3
  48. 48 InternVL3.5-241B-A28B 67.0
  49. 49 Qwen3-VL-8B-Instruct 67.0
  50. 50 MiMo-VL-7B-RL 67.0
  51. 51 Gemini 2.0 Flash Feb 2025 66.8
  52. 52 MiMo-VL-7B-SFT 66.3
  53. 53 InternVL3.5-38B 65.2
  54. 54 Claude Sonnet 4 65.0
  55. 55 InternVL3-78B 64.8
  56. 56 SenseChat-SenseNova-V6-Pro 64.5
  57. 57 Skywork-R1V3-38B 64.3
  58. 58 GLM-4V 64.2
  59. 59 Step-1o-vision-32k 63.4
  60. 60 GPT-4o-1120 62.6
  61. 61 Qwen2.5-VL-32B 62.5
  62. 62 SenseChat-Vision-V5.5 62.4
  63. 63 Keye-VL-1.5-8B thinking 61.6
  64. 64 Llama 4 Scout 61.6
  65. 65 Qwen2.5-72B-VL-Instruct 61.5
  66. 66 Doubao-Pro-32k 61.4
  67. 67 InternVL3.5-30B-A3B 61.4
  68. 68 MiniCPM-V 4.5 61.1
  69. 69 Claude 3.7 Sonnet 60.8
  70. 70 InternVL3.5-14B 60.5
  71. 71 GPT-4o-0513 59.5
  72. 72 InternVL3.5-8B 59.2
  73. 73 Claude-3.5-Sonnet-1022 59.1
  74. 74 Skywork-R1V-38B 58.2
  75. 75 InternVL2_5-78B 58.0
  76. 76 Keye-VL-8B-Preview-Auto 56.8
  77. 77 Keye-VL-8B-Preview thinking 55.9
  78. 78 BLIP3o-8B 55.9
  79. 79 Gemma 3 27B 55.0
  80. 80 InternVL3.5-GPT-OSS-20B-A4B-Preview 54.5
  81. 81 Hunyuan-Vision-20250103 54.0
  82. 82 BAGEL-7B-MoT 53.9
  83. 83 Qwen2.5-VL-7B 53.4
  84. 84 Qwen-Omni-Turbo 52.9
  85. 85 Gemma 3 12B 51.1
  86. 86 Ming-Lite-Omni 49.7
  87. 87 Ming-Lite-Omni-1.5 48.7
  88. 88 Qwen2.5-Omni-3B 45.7
  89. 89 Qwen2.5-Omni-7B 42.4
  90. 90 MiniCPM-o-2_6 41.9
  91. 91 Gemma 3 4B 40.0

Source: Official AGI-Eval Community Image and Video Understanding leaderboard

  1. 1 Gemini 3 Flash 86.5
  2. 2 Qwen3.5-Plus 86.2
  3. 3 Gemini 3.1 Pro 85.9
  4. 4 Kimi K2.6 84.3
  5. 5 Seed1.8 82.1
  6. 6 Qwen3.5-Flash 81.8
  7. 7 Kimi K2.5 81.7
  8. 8 Seed1.5-VL 81.1
  9. 9 Gemini 2.5 Pro Jun 2025 80.1
  10. 10 Qwen3-VL-235B-A22B thinking 79.8
  11. 11 Gemini 3.1 Flash-Lite 78.3
  12. 12 GPT-5 78.2
  13. 13 Claude Opus 4.6 77.3
  14. 14 GLM-4.6V 76.9
  15. 15 Qwen3-VL-235B-A22B-Instruct 76.8
  16. 16 Gemini 2.5 Flash Jun 2025, non-thinking 76.6
  17. 17 o3 76.5
  18. 18 Claude Sonnet 4.6 75.1
  19. 19 Qwen3-VL-32B-Instruct 74.5
  20. 20 o4-mini 73.7
  21. 21 Qwen3-VL-32B-Thinking 73.4
  22. 22 GLM-4.5V 71.9
  23. 23 MiMo-VL-7B-RL-2508 nothinking 71.6
  24. 24 Intern-S1 71.5
  25. 25 Step 3 71.2
  26. 26 Qwen3-Omni-30B-A3B 70.5
  27. 27 Kimi-VL-A3B-Thinking-2506 69.9
  28. 28 MiMo-VL-7B-SFT-2508 69.6
  29. 29 GLM-4.6V-Flash 69.5
  30. 30 GPT-4.1 68.9
  31. 31 Gemini 2.0 Pro 68.8
  32. 32 GLM-4.1V-9B-Thinking 68.3
  33. 33 GPT-4.5 67.9
  34. 34 Claude Opus 4 67.5
  35. 35 Llama 4 Maverick 67.3
  36. 36 Qwen3-VL-8B-Thinking 67.3
  37. 37 InternVL3.5-241B-A28B 67.0
  38. 38 Qwen3-VL-8B-Instruct 67.0
  39. 39 MiMo-VL-7B-RL 67.0
  40. 40 Gemini 2.0 Flash Feb 2025 66.8
  41. 41 MiMo-VL-7B-SFT 66.3
  42. 42 InternVL3.5-38B 65.2
  43. 43 Claude Sonnet 4 65.0
  44. 44 InternVL3-78B 64.8
  45. 45 Skywork-R1V3-38B 64.3
  46. 46 Qwen2.5-VL-32B 62.5
  47. 47 Keye-VL-1.5-8B thinking 61.6
  48. 48 Llama 4 Scout 61.6
  49. 49 InternVL3.5-30B-A3B 61.4
  50. 50 MiniCPM-V 4.5 61.1
  51. 51 Claude 3.7 Sonnet 60.8
  52. 52 InternVL3.5-14B 60.5
  53. 53 InternVL3.5-8B 59.2
  54. 54 Skywork-R1V-38B 58.2
  55. 55 InternVL2_5-78B 58.0
  56. 56 Keye-VL-8B-Preview thinking 55.9
  57. 57 Gemma 3 27B 55.0
  58. 58 BAGEL-7B-MoT 53.9
  59. 59 Qwen2.5-VL-7B 53.4
  60. 60 Gemma 3 12B 51.1
  61. 61 Qwen2.5-Omni-7B 42.4
  62. 62 MiniCPM-o-2_6 41.9
  63. 63 Gemma 3 4B 40.0

Source: Official AGI-Eval Community Image and Video Understanding leaderboard

#EntryScoreSettingReleasedRun onSource
1 Seed-2.0-Pro-260215 86.7 Mar 2026 Official leaderboard
2 Gemini 3 Flash 86.5 Dec 2025 Dec 2025 Official leaderboard
3 Gemini-3-Pro-Preview-Multimodal 86.2 Dec 2025 Official leaderboard
4 Qwen3.5-Plus 86.2 Feb 2026 Mar 2026 Official leaderboard
5 Gemini 3.1 Pro 85.9 Feb 2026 Mar 2026 Official leaderboard
6 Kimi K2.6 84.3 Apr 2026 Jun 2026 Official leaderboard
7 Seed1.8 82.1 Dec 2025 Jan 2026 Official leaderboard
8 Qwen3.5-Flash 81.8 Feb 2026 Mar 2026 Official leaderboard
9 Kimi K2.5 81.7 Jan 2026 Feb 2026 Official leaderboard
10 Seed1.5-VL 81.1 May 2025 Aug 2025 Official leaderboard
11 Gemini 2.5 Pro Jun 2025 80.1 Jun 2025 Jul 2025 Official leaderboard
12 Qwen3-VL-235B-A22B 79.8 thinking Sep 2025 Nov 2025 Official leaderboard
13 Gemini 3.1 Flash-Lite 78.3 Mar 2026 Mar 2026 Official leaderboard
14 GPT-5 78.2 Aug 2025 Sep 2025 Official leaderboard
15 Doubao-Seed-1.6-Thinking-2507 77.7 Sep 2025 Official leaderboard
16 Claude Opus 4.6 77.3 Feb 2026 Mar 2026 Official leaderboard
17 GLM-4.6V 76.9 Dec 2025 Jan 2026 Official leaderboard
18 Qwen3-VL-235B-A22B-Instruct 76.8 Sep 2025 Nov 2025 Official leaderboard
19 Gemini 2.5 Flash Jun 2025 76.6 non-thinking Jun 2025 Sep 2025 Official leaderboard
20 o3 76.5 Apr 2025 Mar 2026 Official leaderboard
21 Doubao-Seed-1.6-Thinking-2506 76.3 Sep 2025 Official leaderboard
22 Doubao-Seed-1.6-2506 75.5 Aug 2025 Official leaderboard
23 Claude Sonnet 4.6 75.1 Feb 2026 Mar 2026 Official leaderboard
24 Doubao-1.5-Vision-Pro 74.8 Aug 2025 Official leaderboard
25 Qwen3-VL-32B-Instruct 74.5 Oct 2025 Dec 2025 Official leaderboard
26 o4-mini 73.7 Apr 2025 Mar 2026 Official leaderboard
27 Doubao-Seed-1.6-Flash-250615 73.6 Nov 2026 Official leaderboard
28 Qwen3-VL-32B-Thinking 73.4 Oct 2025 Dec 2025 Official leaderboard
29 Doubao-1.5-Vision-Pro-32k-250115 73.3 Nov 2026 Official leaderboard
30 GLM-4.5V 71.9 Aug 2025 Aug 2025 Official leaderboard
31 MiMo-VL-7B-RL-2508 71.6 nothinking Aug 2025 Sep 2025 Official leaderboard
32 Intern-S1 71.5 Jul 2025 Sep 2025 Official leaderboard
33 Qwen3-Omni-Flash-2025-1201 71.4 Dec 2025 Official leaderboard
34 Step 3 71.2 Jul 2025 Aug 2025 Official leaderboard
35 Kimi-Thinking-Preview 71.0 Sep 2025 Official leaderboard
36 Qwen3-Omni-30B-A3B 70.5 Sep 2025 Mar 2026 Official leaderboard
37 Kimi-VL-A3B-Thinking-2506 69.9 Jun 2025 Sep 2025 Official leaderboard
38 MiMo-VL-7B-SFT-2508 69.6 Aug 2025 Sep 2025 Official leaderboard
39 GLM-4.6V-Flash 69.5 Dec 2025 Jan 2026 Official leaderboard
40 GPT-4.1 68.9 Apr 2025 Sep 2025 Official leaderboard
41 Gemini 2.0 Pro 68.8 Feb 2025 Sep 2025 Official leaderboard
42 GLM-4.1V-9B-Thinking 68.3 Jul 2025 Apr 2026 Official leaderboard
43 Doubao-1.5-Vision-Lite-250315 68.0 Nov 2026 Official leaderboard
44 GPT-4.5 67.9 Feb 2025 Jul 2025 Official leaderboard
45 Claude Opus 4 67.5 May 2025 Aug 2025 Official leaderboard
46 Llama 4 Maverick 67.3 Apr 2025 Mar 2026 Official leaderboard
47 Qwen3-VL-8B-Thinking 67.3 Oct 2025 Dec 2025 Official leaderboard
48 InternVL3.5-241B-A28B 67.0 Aug 2025 Sep 2025 Official leaderboard
49 Qwen3-VL-8B-Instruct 67.0 Oct 2025 Dec 2025 Official leaderboard
50 MiMo-VL-7B-RL 67.0 May 2025 Aug 2025 Official leaderboard
51 Gemini 2.0 Flash Feb 2025 66.8 Feb 2025 Jun 2025 Official leaderboard
52 MiMo-VL-7B-SFT 66.3 May 2025 Nov 2026 Official leaderboard
53 InternVL3.5-38B 65.2 Aug 2025 Sep 2025 Official leaderboard
54 Claude Sonnet 4 65.0 May 2025 Sep 2025 Official leaderboard
55 InternVL3-78B 64.8 Apr 2025 Nov 2026 Official leaderboard
56 SenseChat-SenseNova-V6-Pro 64.5 Nov 2026 Official leaderboard
57 Skywork-R1V3-38B 64.3 Jul 2025 Aug 2025 Official leaderboard
58 GLM-4V 64.2 May 2025 Official leaderboard
59 Step-1o-vision-32k 63.4 Nov 2026 Official leaderboard
60 GPT-4o-1120 62.6 Mar 2026 Official leaderboard
61 Qwen2.5-VL-32B 62.5 Mar 2025 Sep 2025 Official leaderboard
62 SenseChat-Vision-V5.5 62.4 Jun 2025 Official leaderboard
63 Keye-VL-1.5-8B 61.6 thinking Aug 2025 Sep 2025 Official leaderboard
64 Llama 4 Scout 61.6 Apr 2025 Mar 2026 Official leaderboard
65 Qwen2.5-72B-VL-Instruct 61.5 Nov 2026 Official leaderboard
66 Doubao-Pro-32k 61.4 Nov 2026 Official leaderboard
67 InternVL3.5-30B-A3B 61.4 Aug 2025 Sep 2025 Official leaderboard
68 MiniCPM-V 4.5 61.1 Aug 2025 Sep 2025 Official leaderboard
69 Claude 3.7 Sonnet 60.8 Feb 2025 Mar 2026 Official leaderboard
70 InternVL3.5-14B 60.5 Aug 2025 Sep 2025 Official leaderboard
71 GPT-4o-0513 59.5 Jan 2026 Official leaderboard
72 InternVL3.5-8B 59.2 Aug 2025 Sep 2025 Official leaderboard
73 Claude-3.5-Sonnet-1022 59.1 Mar 2026 Official leaderboard
74 Skywork-R1V-38B 58.2 Mar 2025 May 2025 Official leaderboard
75 InternVL2_5-78B 58.0 Dec 2024 Nov 2026 Official leaderboard
76 Keye-VL-8B-Preview-Auto 56.8 Aug 2025 Official leaderboard
77 Keye-VL-8B-Preview 55.9 thinking Jun 2025 Aug 2025 Official leaderboard
78 BLIP3o-8B 55.9 Sep 2025 Official leaderboard
79 Gemma 3 27B 55.0 Mar 2025 Mar 2026 Official leaderboard
80 InternVL3.5-GPT-OSS-20B-A4B-Preview 54.5 Sep 2025 Official leaderboard
81 Hunyuan-Vision-20250103 54.0 May 2025 Official leaderboard
82 BAGEL-7B-MoT 53.9 May 2025 Aug 2025 Official leaderboard
83 Qwen2.5-VL-7B 53.4 Jan 2025 Nov 2026 Official leaderboard
84 Qwen-Omni-Turbo 52.9 May 2025 Official leaderboard
85 Gemma 3 12B 51.1 Mar 2025 Mar 2026 Official leaderboard
86 Ming-Lite-Omni 49.7 Aug 2025 Official leaderboard
87 Ming-Lite-Omni-1.5 48.7 Aug 2025 Official leaderboard
88 Qwen2.5-Omni-3B 45.7 Aug 2025 Official leaderboard
89 Qwen2.5-Omni-7B 42.4 Mar 2025 Sep 2025 Official leaderboard
90 MiniCPM-o-2_6 41.9 Jan 2025 May 2025 Official leaderboard
91 Gemma 3 4B 40.0 Mar 2025 Feb 2026 Official leaderboard
#ModelScoreSettingReleasedRun onSource
1 Gemini 3 Flash 86.5 Dec 2025 Dec 2025 Official leaderboard
2 Qwen3.5-Plus 86.2 Feb 2026 Mar 2026 Official leaderboard
3 Gemini 3.1 Pro 85.9 Feb 2026 Mar 2026 Official leaderboard
4 Kimi K2.6 84.3 Apr 2026 Jun 2026 Official leaderboard
5 Seed1.8 82.1 Dec 2025 Jan 2026 Official leaderboard
6 Qwen3.5-Flash 81.8 Feb 2026 Mar 2026 Official leaderboard
7 Kimi K2.5 81.7 Jan 2026 Feb 2026 Official leaderboard
8 Seed1.5-VL 81.1 May 2025 Aug 2025 Official leaderboard
9 Gemini 2.5 Pro Jun 2025 80.1 Jun 2025 Jul 2025 Official leaderboard
10 Qwen3-VL-235B-A22B 79.8 thinking Sep 2025 Nov 2025 Official leaderboard
11 Gemini 3.1 Flash-Lite 78.3 Mar 2026 Mar 2026 Official leaderboard
12 GPT-5 78.2 Aug 2025 Sep 2025 Official leaderboard
13 Claude Opus 4.6 77.3 Feb 2026 Mar 2026 Official leaderboard
14 GLM-4.6V 76.9 Dec 2025 Jan 2026 Official leaderboard
15 Qwen3-VL-235B-A22B-Instruct 76.8 Sep 2025 Nov 2025 Official leaderboard
16 Gemini 2.5 Flash Jun 2025 76.6 non-thinking Jun 2025 Sep 2025 Official leaderboard
17 o3 76.5 Apr 2025 Mar 2026 Official leaderboard
18 Claude Sonnet 4.6 75.1 Feb 2026 Mar 2026 Official leaderboard
19 Qwen3-VL-32B-Instruct 74.5 Oct 2025 Dec 2025 Official leaderboard
20 o4-mini 73.7 Apr 2025 Mar 2026 Official leaderboard
21 Qwen3-VL-32B-Thinking 73.4 Oct 2025 Dec 2025 Official leaderboard
22 GLM-4.5V 71.9 Aug 2025 Aug 2025 Official leaderboard
23 MiMo-VL-7B-RL-2508 71.6 nothinking Aug 2025 Sep 2025 Official leaderboard
24 Intern-S1 71.5 Jul 2025 Sep 2025 Official leaderboard
25 Step 3 71.2 Jul 2025 Aug 2025 Official leaderboard
26 Qwen3-Omni-30B-A3B 70.5 Sep 2025 Mar 2026 Official leaderboard
27 Kimi-VL-A3B-Thinking-2506 69.9 Jun 2025 Sep 2025 Official leaderboard
28 MiMo-VL-7B-SFT-2508 69.6 Aug 2025 Sep 2025 Official leaderboard
29 GLM-4.6V-Flash 69.5 Dec 2025 Jan 2026 Official leaderboard
30 GPT-4.1 68.9 Apr 2025 Sep 2025 Official leaderboard
31 Gemini 2.0 Pro 68.8 Feb 2025 Sep 2025 Official leaderboard
32 GLM-4.1V-9B-Thinking 68.3 Jul 2025 Apr 2026 Official leaderboard
33 GPT-4.5 67.9 Feb 2025 Jul 2025 Official leaderboard
34 Claude Opus 4 67.5 May 2025 Aug 2025 Official leaderboard
35 Llama 4 Maverick 67.3 Apr 2025 Mar 2026 Official leaderboard
36 Qwen3-VL-8B-Thinking 67.3 Oct 2025 Dec 2025 Official leaderboard
37 InternVL3.5-241B-A28B 67.0 Aug 2025 Sep 2025 Official leaderboard
38 Qwen3-VL-8B-Instruct 67.0 Oct 2025 Dec 2025 Official leaderboard
39 MiMo-VL-7B-RL 67.0 May 2025 Aug 2025 Official leaderboard
40 Gemini 2.0 Flash Feb 2025 66.8 Feb 2025 Jun 2025 Official leaderboard
41 MiMo-VL-7B-SFT 66.3 May 2025 Nov 2026 Official leaderboard
42 InternVL3.5-38B 65.2 Aug 2025 Sep 2025 Official leaderboard
43 Claude Sonnet 4 65.0 May 2025 Sep 2025 Official leaderboard
44 InternVL3-78B 64.8 Apr 2025 Nov 2026 Official leaderboard
45 Skywork-R1V3-38B 64.3 Jul 2025 Aug 2025 Official leaderboard
46 Qwen2.5-VL-32B 62.5 Mar 2025 Sep 2025 Official leaderboard
47 Keye-VL-1.5-8B 61.6 thinking Aug 2025 Sep 2025 Official leaderboard
48 Llama 4 Scout 61.6 Apr 2025 Mar 2026 Official leaderboard
49 InternVL3.5-30B-A3B 61.4 Aug 2025 Sep 2025 Official leaderboard
50 MiniCPM-V 4.5 61.1 Aug 2025 Sep 2025 Official leaderboard
51 Claude 3.7 Sonnet 60.8 Feb 2025 Mar 2026 Official leaderboard
52 InternVL3.5-14B 60.5 Aug 2025 Sep 2025 Official leaderboard
53 InternVL3.5-8B 59.2 Aug 2025 Sep 2025 Official leaderboard
54 Skywork-R1V-38B 58.2 Mar 2025 May 2025 Official leaderboard
55 InternVL2_5-78B 58.0 Dec 2024 Nov 2026 Official leaderboard
56 Keye-VL-8B-Preview 55.9 thinking Jun 2025 Aug 2025 Official leaderboard
57 Gemma 3 27B 55.0 Mar 2025 Mar 2026 Official leaderboard
58 BAGEL-7B-MoT 53.9 May 2025 Aug 2025 Official leaderboard
59 Qwen2.5-VL-7B 53.4 Jan 2025 Nov 2026 Official leaderboard
60 Gemma 3 12B 51.1 Mar 2025 Mar 2026 Official leaderboard
61 Qwen2.5-Omni-7B 42.4 Mar 2025 Sep 2025 Official leaderboard
62 MiniCPM-o-2_6 41.9 Jan 2025 May 2025 Official leaderboard
63 Gemma 3 4B 40.0 Mar 2025 Feb 2026 Official leaderboard

Rankings by part

We read 11 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Image to text objective accuracy, Multi-image understanding and Character information recognition. Qwen3.5-Plus leads 5 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Image to text objective accuracy

63 models we track have a score here.

  1. 1 Gemini 3 Flash 86.5
  2. 2 Qwen3.5-Plus 86.2
  3. 3 Gemini 3.1 Pro 85.9
  4. 4 Kimi K2.6 84.3
  5. 5 Seed1.8 82.1
  6. 6 Qwen3.5-Flash 81.8
  7. 7 Kimi K2.5 81.7
  8. 8 Seed1.5-VL 81.1
  9. 9 Gemini 2.5 Pro Jun 2025 80.1
  10. 10 Qwen3-VL-235B-A22B 79.8

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Seed1.5-VL
thinking
81.1
MiMo-VL-7B-RL-2508
nothinking
71.6
Keye-VL-1.5-8B
thinking
61.6
Keye-VL-8B-Preview
thinking
55.9

Source: Official AGI-Eval Community Image and Video Understanding leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 3 Flash 86.5
Qwen3.5-Plus 86.2
Gemini 3.1 Pro 85.9
Kimi K2.6 84.3
Seed1.8 82.1
Qwen3.5-Flash 81.8
Kimi K2.5 81.7
Seed1.5-VL 81.1
Seed1.5-VL thinking 80.6
Gemini 2.5 Pro Jun 2025 80.1
Qwen3-VL-235B-A22B thinking 79.8
Gemini 2.5 Pro Jun 2025 78.9
Gemini 3.1 Flash-Lite 78.3
GPT-5 78.2
Claude Opus 4.6 77.3
GLM-4.6V 76.9
Qwen3-VL-235B-A22B-Instruct 76.8
Gemini 2.5 Flash Jun 2025 non-thinking 76.6
o3 76.5
Claude Sonnet 4.6 75.1
Qwen3-VL-32B-Instruct 74.5
o4-mini 73.7
Qwen3-VL-32B-Thinking 73.4
GLM-4.5V 71.9
MiMo-VL-7B-RL-2508 nothinking 71.6
Intern-S1 71.5
MiMo-VL-7B-RL-2508 71.4
Step 3 71.2
Qwen3-Omni-30B-A3B 70.5
Kimi-VL-A3B-Thinking-2506 69.9
MiMo-VL-7B-SFT-2508 69.6
GLM-4.6V-Flash 69.5
GPT-4.1 68.9
Gemini 2.0 Pro 68.8
GLM-4.1V-9B-Thinking 68.3
GPT-4.5 67.9
Claude Opus 4 67.5
Llama 4 Maverick 67.3
Qwen3-VL-8B-Thinking 67.3
InternVL3.5-241B-A28B 67.0
Qwen3-VL-8B-Instruct 67.0
MiMo-VL-7B-RL 67.0
Gemini 2.0 Flash Feb 2025 66.8
MiMo-VL-7B-SFT 66.3
InternVL3.5-38B 65.2
Claude Sonnet 4 65.0
InternVL3-78B 64.8
Skywork-R1V3-38B 64.3
Qwen2.5-VL-32B 62.5
Keye-VL-1.5-8B thinking 61.6
Llama 4 Scout 61.6
InternVL3.5-30B-A3B 61.4
MiniCPM-V 4.5 61.1
Claude 3.7 Sonnet 60.8
InternVL3.5-14B 60.5
InternVL3.5-8B 59.2
Skywork-R1V-38B 58.2
InternVL2_5-78B 58.0
Keye-VL-8B-Preview thinking 55.9
Keye-VL-8B-Preview 55.7
Gemma 3 27B 55.0
BAGEL-7B-MoT 53.9
Qwen2.5-VL-7B 53.4
Keye-VL-1.5-8B 52.1
Gemma 3 12B 51.1
Qwen2.5-Omni-7B 42.4
MiniCPM-o-2_6 41.9
Gemma 3 4B 40.0

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
Seed1.5-VL
thinking
81.1
MiMo-VL-7B-RL-2508
nothinking
71.6
Keye-VL-1.5-8B
thinking
61.6
Keye-VL-8B-Preview
thinking
55.9

Source: Official AGI-Eval Community Image and Video Understanding leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 3 Flash 86.5
Qwen3.5-Plus 86.2
Gemini 3.1 Pro 85.9
Kimi K2.6 84.3
Seed1.8 82.1
Qwen3.5-Flash 81.8
Kimi K2.5 81.7
Seed1.5-VL 81.1
Seed1.5-VL thinking 80.6
Gemini 2.5 Pro Jun 2025 80.1
Qwen3-VL-235B-A22B thinking 79.8
Gemini 2.5 Pro Jun 2025 78.9
Gemini 3.1 Flash-Lite 78.3
GPT-5 78.2
Claude Opus 4.6 77.3
GLM-4.6V 76.9
Qwen3-VL-235B-A22B-Instruct 76.8
Gemini 2.5 Flash Jun 2025 non-thinking 76.6
o3 76.5
Claude Sonnet 4.6 75.1
Qwen3-VL-32B-Instruct 74.5
o4-mini 73.7
Qwen3-VL-32B-Thinking 73.4
GLM-4.5V 71.9
MiMo-VL-7B-RL-2508 nothinking 71.6
Intern-S1 71.5
MiMo-VL-7B-RL-2508 71.4
Step 3 71.2
Qwen3-Omni-30B-A3B 70.5
Kimi-VL-A3B-Thinking-2506 69.9
MiMo-VL-7B-SFT-2508 69.6
GLM-4.6V-Flash 69.5
GPT-4.1 68.9
Gemini 2.0 Pro 68.8
GLM-4.1V-9B-Thinking 68.3
GPT-4.5 67.9
Claude Opus 4 67.5
Llama 4 Maverick 67.3
Qwen3-VL-8B-Thinking 67.3
InternVL3.5-241B-A28B 67.0
Qwen3-VL-8B-Instruct 67.0
MiMo-VL-7B-RL 67.0
Gemini 2.0 Flash Feb 2025 66.8
MiMo-VL-7B-SFT 66.3
InternVL3.5-38B 65.2
Claude Sonnet 4 65.0
InternVL3-78B 64.8
Skywork-R1V3-38B 64.3
Qwen2.5-VL-32B 62.5
Keye-VL-1.5-8B thinking 61.6
Llama 4 Scout 61.6
InternVL3.5-30B-A3B 61.4
MiniCPM-V 4.5 61.1
Claude 3.7 Sonnet 60.8
InternVL3.5-14B 60.5
InternVL3.5-8B 59.2
Skywork-R1V-38B 58.2
InternVL2_5-78B 58.0
Keye-VL-8B-Preview thinking 55.9
Keye-VL-8B-Preview 55.7
Gemma 3 27B 55.0
BAGEL-7B-MoT 53.9
Qwen2.5-VL-7B 53.4
Keye-VL-1.5-8B 52.1
Gemma 3 12B 51.1
Qwen2.5-Omni-7B 42.4
MiniCPM-o-2_6 41.9
Gemma 3 4B 40.0

6 models have held number one on the official leaderboard since its release in February 2025

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

Shanghai AI Lab Google DeepMind OpenAI ByteDance Seed
Own leaderboard Not tested since Apr 2026

Source: Official AGI-Eval Community Image and Video Understanding leaderboard

Scores over time

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
40.0 60.0 80.0 2025 2026 40.0 60.0 80.0 2025 2026

Source: Official AGI-Eval Community Image and Video Understanding 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 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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 7 months ago, in April 2026, for Kimi K2.6. Every benchmark

Catalog record 2 pass
  • Pass Task access The tasks are private.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (91 entries).
  • Pass Models scored 63 models we track, plus 28 other entries on its leaderboard.
  • Warning Newest score The newest came 7 months ago, in April 2026, for Kimi K2.6.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude Opus 4.6 (#39).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on agi-eval.org: 68 rows for models we track and 28 other entries at the last read.
  • Pass Last read November 21st, 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