Benchmark Almanac

BrowseComp

Claude Fable 5 93.3% leads since June 2026
In use

Hard-to-find facts that need persistent web browsing. Short verifiable answers graded for exact correctness.

Questions ship encrypted in simple-evals. Context-management settings (compaction at 300K vs full 1M) change scores by several points.

Made by
OpenAI
Released
April 2025
Kind
Web browsing
Level
Expert
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Strongest model tested
GPT-6 Astra #3 on the capability index
Last in a launch post
November 2026
Last model added
November 2026

In the authors’ words

We present BrowseComp, a simple yet challenging benchmark for measuring the ability for agents to browse the web. BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information. Despite the difficulty of the questions, BrowseComp is simple and easy-to-use, as predicted answers are short and easily verifiable against reference answers. BrowseComp for browsing agents can be seen as analogous to how programming competitions are an incomplete but useful benchmark for coding agents. While BrowseComp sidesteps challenges of a true user query distribution, like generating long answers or resolving ambiguity, it measures the important core capability of exercising persistence and creativity in finding information. BrowseComp can be found at https://github.com/openai/simple-evals.

Abstract of BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents (arXiv:2504.12516, CC0)
81
Models scored in lab cards
78
Launch posts list it, from 23 labs
34
System cards report it, from 13 labs

Claude Fable 5 leads with 93.3%

Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.

81 models. Scroll sideways for the rest.

  1. 93.3 Claude Fable 5
  2. 92.2 GPT-5.6 Sol
  3. 92.1 Kimi K3.1 Pro
  4. 91.9 LongCat-2.5
  5. 91.7 Ling-3.1-flash
  6. 91.5 GPT-6 Astra
  7. 91.2 Kimi K3
  8. 90.8 Claude Opus 5
  9. 90.2 Inkling 2 Small
  10. 90.1 GPT-5.5
  11. 90.1 GPT-5.5 Pro
  12. 89.3 GPT-5.4 Pro
  13. 88.7 Step 5 Preview
  14. 88.5 Claude Opus 4.8
  15. 87.5 GPT-5.6 Terra
  16. 86.9 Claude Mythos Preview
  17. 86.6 Claude Sonnet 5
  18. 86.2 Seed2.1 Pro
  19. 85.9 Gemini 3.1 Pro
  20. 84.5 Mistral Large 4 (Le Chonk, open weights)
  21. 84.2 Tencent Hy3
  22. 84.0 Claude Opus 4.6
  23. 83.5 MiniMax-M3
  24. 83.4 DeepSeek-V4-Pro
  25. 83.3 GPT-5.6 Luna
  26. 83.2 Kimi K2.6
  27. 82.7 GPT-5.4
  28. 79.9 LongCat-2.0
  29. 79.3 GLM-5.1
  30. 79.3 Claude Opus 4.7
  31. 78.9 Nemotron 3.5 Super VL
  32. 78.6 Qwen3.5 397B-A17B
  33. 77.9 GPT-5.2 Pro
  34. 77.8 MiniMax-M2.7
  35. 77.4 Inkling-Small
  36. 77.4 Beam 1.0
  37. 77.3 Seed2.0 Pro
  38. 77.1 Inkling
  39. 76.3 MiniMax-M2.5
  40. 75.9 GLM-5
  41. 75.8 Step 3.7 Flash
  42. 74.9 Kimi K2.5
  43. 74.7 Claude Sonnet 4.6
  44. 73.2 DeepSeek-V4-Flash
  45. 73.1 LongCat-Flash-Thinking-2601
  46. 69.0 Step 3.5 Flash
  47. 68.9 ChatGPT agent
  48. 67.6 DeepSeek-V3.2
  49. 67.6 Seed1.8
  50. 67.5 GLM-4.7
  51. 67.1 Tencent Hy3 preview
  52. 66.8 Nemotron 3.5 Bolt 74B-A9B
  53. 64.0 Seed2.0 Lite (0428)
  54. 63.8 Qwen3.5-122B-A10B
  55. 62.0 MiniMax-M2.1
  56. 61.0 Qwen3.5-27B
  57. 61.0 Qwen3.5-35B-A3B
  58. 60.2 Kimi K2 Thinking
  59. 58.3 MiMo-V2-Flash
  60. 54.9 GPT-5
  61. 53.1 Spark-X2.6-4B
  62. 49.7 o3
  63. 49.5 Sarvam 105B
  64. 49.2 Solar Pro 4
  65. 48.6 Mistral Medium 3.5
  66. 45.1 GLM-4.6
  67. 44.9 Grok 4 Fast
  68. 44.4 Nemotron 3 Ultra
  69. 44.0 MiniMax-M2
  70. 42.8 glm-4.7-flash
  71. 40.1 DeepSeek-V3.2-Exp
  72. 39.5 Nemotron 3.5 Lightning 30B-A3B
  73. 38.5 DeepSeek-V3.1-Terminus
  74. 35.5 Sarvam 30B
  75. 31.4 K-EXAONE
  76. 31.3 Nemotron 3 Super
  77. 30.0 DeepSeek-V3.1
  78. 29.6 UI-TARS-2
  79. 28.3 o4-mini
  80. 26.4 GLM-4.5
  81. 21.3 GLM-4.5-Air
  1. 1 Claude Fable 5 93.3%
  2. 2 GPT-5.6 Sol 92.2%
  3. 3 Kimi K3.1 Pro 92.1%
  4. 4 LongCat-2.5 91.9%
  5. 5 Ling-3.1-flash 91.7%
  6. 6 GPT-6 Astra 91.5%
  7. 7 Kimi K3 91.2%
  8. 8 Claude Opus 5 90.8%
  9. 9 Inkling 2 Small 90.2%
  10. 10 GPT-5.5 90.1%
  11. 11 GPT-5.5 Pro 90.1%
  12. 12 GPT-5.4 Pro 89.3%
  13. 13 Step 5 Preview 88.7%
  14. 14 Claude Opus 4.8 88.5%
  15. 15 GPT-5.6 Terra 87.5%
  16. 16 Claude Mythos Preview 86.9%
  17. 17 Claude Sonnet 5 86.6%
  18. 18 Seed2.1 Pro 86.2%
  19. 19 Gemini 3.1 Pro 85.9%
  20. 20 Mistral Large 4 (Le Chonk, open weights) 84.5%
  21. 21 Tencent Hy3 84.2%
  22. 22 Claude Opus 4.6 84%
  23. 23 MiniMax-M3 83.5%
  24. 24 DeepSeek-V4-Pro 83.4%
  25. 25 GPT-5.6 Luna 83.3%
  26. 26 Kimi K2.6 83.2%
  27. 27 GPT-5.4 82.7%
  28. 28 LongCat-2.0 79.9%
  29. 29 GLM-5.1 79.3%
  30. 30 Claude Opus 4.7 79.3%
  31. 31 Nemotron 3.5 Super VL 78.9%
  32. 32 Qwen3.5 397B-A17B 78.6%
  33. 33 GPT-5.2 Pro 77.9%
  34. 34 MiniMax-M2.7 77.8%
  35. 35 Inkling-Small 77.4%
  36. 36 Beam 1.0 77.4%
  37. 37 Seed2.0 Pro 77.3%
  38. 38 Inkling 77.1%
  39. 39 MiniMax-M2.5 76.3%
  40. 40 GLM-5 75.9%
  41. 41 Step 3.7 Flash 75.8%
  42. 42 Kimi K2.5 74.9%
  43. 43 Claude Sonnet 4.6 74.7%
  44. 44 DeepSeek-V4-Flash 73.2%
  45. 45 LongCat-Flash-Thinking-2601 73.1%
  46. 46 Step 3.5 Flash 69%
  47. 47 ChatGPT agent 68.9%
  48. 48 DeepSeek-V3.2 67.6%
  49. 49 Seed1.8 67.6%
  50. 50 GLM-4.7 67.5%
  51. 51 Tencent Hy3 preview 67.1%
  52. 52 Nemotron 3.5 Bolt 74B-A9B 66.8%
  53. 53 Seed2.0 Lite (0428) 64%
  54. 54 Qwen3.5-122B-A10B 63.8%
  55. 55 MiniMax-M2.1 62%
  56. 56 Qwen3.5-27B 61%
  57. 57 Qwen3.5-35B-A3B 61%
  58. 58 Kimi K2 Thinking 60.2%
  59. 59 MiMo-V2-Flash 58.3%
  60. 60 GPT-5 54.9%
  61. 61 Spark-X2.6-4B 53.1%
  62. 62 o3 49.7%
  63. 63 Sarvam 105B 49.5%
  64. 64 Solar Pro 4 49.2%
  65. 65 Mistral Medium 3.5 48.6%
  66. 66 GLM-4.6 45.1%
  67. 67 Grok 4 Fast 44.9%
  68. 68 Nemotron 3 Ultra 44.4%
  69. 69 MiniMax-M2 44%
  70. 70 glm-4.7-flash 42.8%
  71. 71 DeepSeek-V3.2-Exp 40.1%
  72. 72 Nemotron 3.5 Lightning 30B-A3B 39.5%
  73. 73 DeepSeek-V3.1-Terminus 38.5%
  74. 74 Sarvam 30B 35.5%
  75. 75 K-EXAONE 31.4%
  76. 76 Nemotron 3 Super 31.3%
  77. 77 DeepSeek-V3.1 30%
  78. 78 UI-TARS-2 29.6%
  79. 79 o4-mini 28.3%
  80. 80 GLM-4.5 26.4%
  81. 81 GLM-4.5-Air 21.3%

Source: Labs’ model cards

#ModelScoreSettingDateSource
1 Claude Fable 5 93.3% Jun 2026 Lab model card
2 GPT-5.6 Sol 92.2% Jul 2026 Lab model card
3 Kimi K3.1 Pro 92.1% Oct 2026 Lab model card
4 LongCat-2.5 91.9% Nov 2026 Lab model card
5 Ling-3.1-flash 91.7% Oct 2026 Lab model card
6 GPT-6 Astra 91.5% Sep 2026 Lab model card
7 Kimi K3 91.2% Jul 2026 Lab model card
8 Claude Opus 5 90.8% Jul 2026 Lab model card
9 Inkling 2 Small 90.2% Nov 2026 Lab model card
10 GPT-5.5 90.1% Apr 2026 Lab model card
11 GPT-5.5 Pro 90.1% Apr 2026 Lab model card
12 GPT-5.4 Pro 89.3% Mar 2026 Lab model card
13 Step 5 Preview 88.7% Sep 2026 Lab model card
14 Claude Opus 4.8 88.5% May 2026 Lab model card
15 GPT-5.6 Terra 87.5% Jul 2026 Lab model card
16 Claude Mythos Preview 86.9% Apr 2026 Lab model card
17 Claude Sonnet 5 86.6% Jun 2026 Lab model card
18 Seed2.1 Pro 86.2% Jun 2026 Lab model card
19 Gemini 3.1 Pro 85.9% Feb 2026 Lab model card
20 Mistral Large 4 (Le Chonk, open weights) 84.5% Oct 2026 Lab model card
21 Tencent Hy3 84.2% Jul 2026 Lab model card
22 Claude Opus 4.6 84% Feb 2026 Lab model card
23 MiniMax-M3 83.5% Jun 2026 Lab model card
24 DeepSeek-V4-Pro 83.4% Apr 2026 Lab model card
25 GPT-5.6 Luna 83.3% Jul 2026 Lab model card
26 Kimi K2.6 83.2% Apr 2026 Lab model card
27 GPT-5.4 82.7% Mar 2026 Lab model card
28 LongCat-2.0 79.9% Jul 2026 Lab model card
29 GLM-5.1 79.3% Apr 2026 Lab model card
30 Claude Opus 4.7 79.3% Apr 2026 Lab model card
31 Nemotron 3.5 Super VL 78.9% Oct 2026 Lab model card
32 Qwen3.5 397B-A17B 78.6% Feb 2026 Lab model card
33 GPT-5.2 Pro 77.9% Dec 2025 Lab model card
34 MiniMax-M2.7 77.8% May 2026 Lab model card
35 Inkling-Small 77.4% Jul 2026 Lab model card
36 Beam 1.0 77.4% Oct 2026 Lab model card
37 Seed2.0 Pro 77.3% Feb 2026 Lab model card
38 Inkling 77.1% Jul 2026 Lab model card
39 MiniMax-M2.5 76.3% Feb 2026 Lab model card
40 GLM-5 75.9% Feb 2026 Lab model card
41 Step 3.7 Flash 75.8% May 2026 Lab model card
42 Kimi K2.5 74.9% Jan 2026 Lab model card
43 Claude Sonnet 4.6 74.7% Feb 2026 Lab model card
44 DeepSeek-V4-Flash 73.2% Apr 2026 Lab model card
45 LongCat-Flash-Thinking-2601 73.1% Jan 2026 Lab model card
46 Step 3.5 Flash 69% Feb 2026 Lab model card
47 ChatGPT agent 68.9% Jul 2025 Lab model card
48 DeepSeek-V3.2 67.6% Dec 2025 Lab model card
49 Seed1.8 67.6% Dec 2025 Lab model card
50 GLM-4.7 67.5% Dec 2025 Lab model card
51 Tencent Hy3 preview 67.1% Apr 2026 Lab model card
52 Nemotron 3.5 Bolt 74B-A9B 66.8% Oct 2026 Lab model card
53 Seed2.0 Lite (0428) 64% Apr 2026 Lab model card
54 Qwen3.5-122B-A10B 63.8% Feb 2026 Lab model card
55 MiniMax-M2.1 62% Dec 2025 Lab model card
56 Qwen3.5-27B 61% Feb 2026 Lab model card
57 Qwen3.5-35B-A3B 61% Feb 2026 Lab model card
58 Kimi K2 Thinking 60.2% Nov 2025 Lab model card
59 MiMo-V2-Flash 58.3% Dec 2025 Lab model card
60 GPT-5 54.9% Aug 2025 Lab model card
61 Spark-X2.6-4B 53.1% Oct 2026 Lab model card
62 o3 49.7% Apr 2025 Lab model card
63 Sarvam 105B 49.5% Mar 2026 Lab model card
64 Solar Pro 4 49.2% Aug 2026 Lab model card
65 Mistral Medium 3.5 48.6% May 2026 Lab model card
66 GLM-4.6 45.1% Sep 2025 Lab model card
67 Grok 4 Fast 44.9% Sep 2025 Lab model card
68 Nemotron 3 Ultra 44.4% Jun 2026 Lab model card
69 MiniMax-M2 44% Oct 2025 Lab model card
70 glm-4.7-flash 42.8% Jan 2026 Lab model card
71 DeepSeek-V3.2-Exp 40.1% Sep 2025 Lab model card
72 Nemotron 3.5 Lightning 30B-A3B 39.5% Oct 2026 Lab model card
73 DeepSeek-V3.1-Terminus 38.5% Sep 2025 Lab model card
74 Sarvam 30B 35.5% Mar 2026 Lab model card
75 K-EXAONE 31.4% Dec 2025 Lab model card
76 Nemotron 3 Super 31.3% Mar 2026 Lab model card
77 DeepSeek-V3.1 30% Aug 2025 Lab model card
78 UI-TARS-2 29.6% Sep 2025 Lab model card
79 o4-mini 28.3% Apr 2025 Lab model card
80 GLM-4.5 26.4% Jul 2025 Lab model card
81 GLM-4.5-Air 21.3% Aug 2025 Lab model card

8 models have held number one in labs’ own numbers since its release in April 2025

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

OpenAI Anthropic Google DeepMind
Lab claims Passed 85% in Feb 2026, by Gemini 3.1 Pro
Claude Fable 5 93.3%

Source: Labs’ model cards

Google DeepMind’s own number for Gemini 3.1 Pro passed 85% in February 2026

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

Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Claude Fable 5 Gemini 3.1 Pro GPT-5.2 Pro ChatGPT agent o3 0% 20% 40% 60% 80% 100% 2025 2026 2027 Near the ceiling (85%) Claude Fable 5 Gemini 3.1 Pro

Source: Labs’ model cards

14 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.

OpenAI
Z.ai
DeepSeek
ByteDance Seed
xAI
MiniMax
Moonshot AI
Xiaomi Corp
LG AI Research
Meituan
StepFun
Anthropic
Qwen
Google DeepMind
Sarvam AI
NVIDIA
Tencent
Mistral AI
Thinking Machines
Upstage
Reflection
Ant Group
iFlytek
DateLaunch postScoreAs the post writes it
Nov 5, 2026 Inkling 2 Small 90.2 BrowseComp with context management, 90.2%
Nov 2, 2026 LongCat-2.5 91.9 BrowseComp
Oct 29, 2026 Mistral Large 4 (Le Chonk, open weights) 84.5 Web browsing BrowseComp
Oct 24, 2026 Kimi K3.1 Pro 92.1 BrowseComp
Oct 22, 2026 Spark-X2.6-4B 53.1 BrowseComp
Oct 15, 2026 Nemotron 3.5 Bolt 74B-A9B 66.8 BrowseComp
Oct 15, 2026 Nemotron 3.5 Super VL 78 BrowseComp (78.91 in the technical report)
Oct 13, 2026 Ling-3.1-flash 91.7 Browsecomp
Oct 5, 2026 Beam 1.0 77.4 BrowseComp w/ context management
Sep 20, 2026 Step 5 Preview 88.7 BrowseComp
Sep 3, 2026 GPT-6 Astra 91.5 BrowseComp
Aug 11, 2026 Nemotron 3.5 Lightning 30B-A3B 37 BrowseComp
Aug 11, 2026 Solar Pro 4 49.2 BrowseComp* (in-house)
Jul 30, 2026 Inkling-Small 77.4 BrowseComp with context management
Jul 24, 2026 Claude Opus 5 90.8 Agentic search BrowseComp
Jul 16, 2026 Kimi K3 91.2 BrowseComp
Jul 15, 2026 Inkling 77.1 BrowseComp w/ ctx management
Jul 9, 2026 GPT-5.6 Luna 83.3 BrowseComp
Jul 9, 2026 GPT-5.6 Terra 87.5 BrowseComp
Jul 9, 2026 GPT-5.6 Sol 92.2 BrowseComp (Sol Ultra, 4 agents)
Jul 5, 2026 LongCat-2.0 79.9 BrowseComp
Jul 2, 2026 Tencent Hy3 84.2 BrowseComp
Jun 30, 2026 Claude Sonnet 5 86.6 BrowseComp (multi agent)
Jun 9, 2026 Claude Fable 5 93.3 BrowseComp (multi-agent)
Jun 4, 2026 Nemotron 3 Ultra 44.4 BrowseComp
Jun 1, 2026 MiniMax-M3 83.5 BrowseComp
May 29, 2026 Step 3.7 Flash 75.8 BrowseComp (acc)
May 28, 2026 Claude Opus 4.8 88.5 BrowseComp (multi-agent)
May 22, 2026 Mistral Medium 3.5 48.6 BrowseComp**** (self-reported, context management, discard-all at 100k tokens)
Apr 28, 2026 Seed2.0 Lite (0428) 64 BrowseComp
Apr 24, 2026 DeepSeek-V4-Flash 73.2 BrowseComp (Pass@1), V4-Flash Max
Apr 24, 2026 DeepSeek-V4-Pro 83.4 BrowseComp (Pass@1)
Apr 23, 2026 GPT-5.5 Pro 90.1 BrowseComp
Apr 23, 2026 Tencent Hy3 preview 67.1 BrowseComp
Apr 23, 2026 GPT-5.5 90.1 BrowseComp (GPT-5.5 Pro)
Apr 20, 2026 Kimi K2.6 83.2 BrowseComp
Apr 16, 2026 Claude Opus 4.7 79.3 Agentic search BrowseComp
Apr 7, 2026 GLM-5.1 79.3 BrowseComp (w/ Context Manage)
Apr 7, 2026 Claude Mythos Preview 86.9 BrowseComp (Glasswing page)
Mar 11, 2026 Nemotron 3 Super 31.3 BrowseComp with Search
Mar 6, 2026 Sarvam 30B 35.5 BrowseComp
Mar 6, 2026 Sarvam 105B 49.5 BrowseComp
Mar 5, 2026 GPT-5.4 Pro 89.3 BrowseComp
Mar 5, 2026 GPT-5.4 82.7 BrowseComp
Feb 24, 2026 Qwen3.5-35B-A3B 61 Browsecomp
Feb 24, 2026 Qwen3.5-27B 61 Browsecomp
Feb 24, 2026 Qwen3.5-122B-A10B 63.8 Browsecomp (context folding)
Feb 19, 2026 Gemini 3.1 Pro 85.9 BrowseComp (Search + Python + Browse)
Feb 17, 2026 Claude Sonnet 4.6 74.7 Agentic search BrowseComp
Feb 16, 2026 Qwen3.5 397B-A17B 78.6 BrowseComp (69.0/78.6)
Feb 14, 2026 Seed2.0 Pro 77.3 BrowseComp
Feb 12, 2026 MiniMax-M2.5 76.3 BrowseComp (w/ctx)
Feb 11, 2026 GLM-5 75.9 BrowseComp (w/ Context Manage)
Feb 5, 2026 Claude Opus 4.6 84 Agentic search BrowseComp
Feb 2, 2026 Step 3.5 Flash 69 Best of BrowseComp 51.6 / BrowseComp (w/ Context Manager) 69.0
Jan 27, 2026 Kimi K2.5 74.9 BrowseComp (w/ctx manage)
Jan 19, 2026 glm-4.7-flash 42.8 BrowseComp
Jan 14, 2026 LongCat-Flash-Thinking-2601 73.1 Best of BrowseComp (Pass@1) 56.6 / with context management 73.1
Dec 31, 2025 K-EXAONE 31.4 BrowseComp
Dec 23, 2025 MiniMax-M2.1 62 BrowseComp (context management)
Dec 22, 2025 GLM-4.7 67.5 BrowseComp (w/ Context Manage)
Dec 18, 2025 Seed1.8 67.6 BrowseComp-en
Dec 16, 2025 MiMo-V2-Flash 58.3 Best of BrowseComp 45.4 / BrowseComp (w/ Context Manage) 58.3
Dec 11, 2025 GPT-5.2 Pro 77.9 BrowseComp
Dec 1, 2025 DeepSeek-V3.2 67.6 BrowseComp (Pass@1)
Nov 6, 2025 Kimi K2 Thinking 60.2 BrowseComp, w/ tools
Oct 27, 2025 MiniMax-M2 44 BrowseComp
Sep 30, 2025 GLM-4.6 45.1 BrowseComp
Sep 29, 2025 DeepSeek-V3.2-Exp 40.1 BrowseComp
Sep 22, 2025 DeepSeek-V3.1-Terminus 38.5 BrowseComp
Sep 19, 2025 Grok 4 Fast 44.9 BrowseComp pass@1
Sep 2, 2025 UI-TARS-2 29.6 BC-en 7.0 GUI-only (29.6 with GUI-SDK)
Aug 21, 2025 DeepSeek-V3.1 30 BrowseComp
Aug 7, 2025 GPT-5 54.9 BrowseComp Agentic search & browsing
Jul 28, 2025 GLM-4.5 26.4 BrowseComp
Jul 17, 2025 ChatGPT agent 68.9 BrowseComp (Agentic Browsing)
Apr 16, 2025 o4-mini 28.3 BrowseComp (with python + browsing)
Apr 16, 2025 o3 49.7 BrowseComp Agentic Browsing, o3 with python + browsing

34 system cards report it, from 13 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
Oct 15, 2026 Nemotron 3.5 Super VL technical report 78.9 BrowseComp, NeMo Gym harness with Exa search (Table 10). OpenCode harness gives 72.35 Capabilities
Oct 15, 2026 Nemotron 3.5 Lightning 30B-A3B technical report 39.5 BrowseComp, BF16 acc. (Table 22) Capabilities
Oct 15, 2026 Nemotron 3.5 Bolt 74B-A9B technical report 66.8 BrowseComp, Exa web search, released 24GB checkpoint (Table 7 and Section 8). BF16 70.00 Capabilities
Jul 30, 2026 Inkling-Small model card 77.4 BrowseComp with context management (5. Evaluations table) Capabilities
Jul 27, 2026 Kimi K3 technical report 91.2 BrowseComp with context compaction at 300K tokens, 90.4 with full 1M context and no context management (Table 2, 6.1.3) Capabilities
Jul 24, 2026 Claude Opus 5 system card 90.8 BrowseComp, single agent with web search, fetch, programmatic tool calling, code execution and compaction (Table 8.1.A, 8.10.2). Multi-agent 10-agent team 93.6% on a pre-release configuration (8.11.1) Capabilities
Jul 15, 2026 Inkling model card 77.1 BrowseComp w/ ctx management (5. Evaluations table, effort 0.99) Capabilities
Jun 30, 2026 Claude Sonnet 5 system card 84.7 BrowseComp single agent, max effort, 10M-token limit with compaction (multi agent 86.6%) (Table 8.1.A, 8.9.2) Capabilities
Jun 23, 2026 Seed2.1 Pro model card 86.2 BrowseComp (with Search) (Table 11) Capabilities
Jun 9, 2026 Nemotron 3 Ultra technical report 44.4 BrowseComp (Table 10) Capabilities
May 28, 2026 Claude Opus 4.8 system card 84.3 BrowseComp single-agent, max effort, 10M-token limit with compaction at 200k (multi-agent 88.5% with an orchestrator and blocking subagents, five-agent team 85.4%) (Table 8.1.A, 8.10.2, 8.11.1) Capabilities
May 26, 2026 MiniMax-M2 technical report 44 BrowseComp (Hugging Face official score), MiniMax-M2 bar, printed label (Figure 9) Capabilities
May 26, 2026 MiniMax-M2.7 technical report 77.8 BrowseComp, WebExplorer framework with context dropping (Table 4) Capabilities
May 26, 2026 MiniMax-M2.5 technical report 76.3 BrowseComp, WebExplorer framework with context dropping (Table 4) Capabilities
Apr 26, 2026 DeepSeek-V4-Pro technical report 83.4 BrowseComp (Pass@1), discard-all context management, DS-V4-Pro Max (Table 6) Capabilities
Apr 26, 2026 DeepSeek-V4-Flash technical report 73.2 BrowseComp (Pass@1), DeepSeek-V4-Flash Max (Table 7) Capabilities
Apr 16, 2026 Claude Opus 4.7 system card 79.3 BrowseComp, thinking off at max effort, 10M token limit with compaction (Table 8.1.A, 8.8.2) Capabilities
Apr 7, 2026 Claude Mythos Preview system card 86.9 BrowseComp, web search, web fetch, programmatic tool calling and code execution, adaptive thinking, max effort, 3M token limit with compaction (6.10.2); 24.0 with no thinking and no tools as a contamination check Capabilities
Mar 21, 2026 Seed1.8 model card 67.6 BrowseComp-en (Table 4). 45.0 low and 55.0 medium effort (Figure 4 text) Capabilities
Mar 11, 2026 Nemotron 3 Super technical report 31.3 BrowseComp with Search, internal harness with Serp API (Table 5) Capabilities
Feb 19, 2026 Gemini 3.1 Pro model card 85.9 BrowseComp, Search + Python + Browse (results table, p.4) Capabilities
Feb 17, 2026 GLM-5 technical report 75.9 BrowseComp (w/ Context Manage), discard-all (Table 7, Figure 1 headline). Without context management 62.0 Capabilities
Feb 17, 2026 Claude Sonnet 4.6 system card 74 BrowseComp single agent, no thinking, max effort, compaction up to 10M tokens, revised 2026-03-06 from 74.72; multi-agent 82.07 (2.20.1.1, 2.20.1.3) Capabilities
Feb 14, 2026 Seed2.0 Pro model card 77.3 BrowseComp, search agent (Table 11) Capabilities
Feb 11, 2026 Step 3.5 Flash technical report 69 BrowseComp (w. Ctx Manage), discard-all context management, as in the abstract (Table 5). Without context management 51.6 Capabilities
Feb 5, 2026 Claude Opus 4.6 system card 83.7 BrowseComp, best single-agent configuration, no thinking, max effort, compaction up to 10M tokens, revised on 2026-03-06 from 83.97 (Figure 2.21.1.A); multi-agent 86.57, revised from 86.81 (2.21.1.2) Capabilities
Jan 30, 2026 Kimi K2.5 technical report 60.6 BrowseComp without context management. 74.9 with discard-all context management, 78.4 with Agent Swarm (Table 4) Capabilities
Jan 23, 2026 LongCat-Flash-Thinking-2601 technical report 73.1 BrowseComp (Pass@1), with context management (Table 2). Without context management 56.6 Capabilities
Jan 6, 2026 MiMo-V2-Flash technical report 58.3 BrowseComp (w/ Context Manage) 58.3, BrowseComp without it 45.4 (Table 9) Capabilities
Jan 5, 2026 K-EXAONE technical report 31.4 BrowseComp, marked Non-reasoning, with summarizer and trajectory compressor (Table 3 and C.6) Capabilities
Dec 1, 2025 DeepSeek-V3.2 technical report 67.6 BrowseComp (Pass@1) with discard-all context management, the final score per 4.1. 51.4 without context management, 60.2 with Summary DeepSeek-V3.2 Thinking (Table 2) Capabilities
Sep 29, 2025 DeepSeek-V3.2-Exp technical report 40.1 BrowseComp (Acc.), search agent, DeepSeek-V3.2-Exp (Table 1) Capabilities
Aug 8, 2025 GLM-4.5 technical report 26.4 BrowseComp (Table 3) Capabilities
Aug 8, 2025 GLM-4.5-Air technical report 21.3 BrowseComp (Table 3) Capabilities

More in Computer and browser use

All 69 in Computer and browser use
  • ScreenSpot-Pro In use GUI grounding on high-resolution professional-software screenshots. The model must locate the target UI element. GPT-6 Astra 92.7%
  • AndroidWorld In use Agents complete 116 tasks across 20 real Android apps in an emulator. Each task checks the device's system state to judge success. GPT-6 Astra 100%
  • OSWorld 2.1 In use Agents complete 108 long-horizon workflows across 31 self-hosted websites and desktop apps, averaging 318 tool calls. Scored by binary completion and checkpoint partial credit. Claude Opus 5.5 81.8%
  • WebArena-Verified In use Agents complete 812 audited WebArena tasks on self-hosted websites. Deterministic evaluators check the agent's answer and captured network traces. Muse Spark 1.1 69%
  • MobileWorld In use Agents complete 201 tasks across 20 mobile apps, including tasks that need user clarification or MCP tools. Backend checks verify success. GPT-5 55.3%
  • RecreationBench In use Agents explore a running reference app and rebuild it, across 250 tasks on five platforms. Programmatic tests and a VLM judge score the result. GPT-6 Astra 58.1%