Benchmark Almanac

Parallel Search Capability Leaderboard

Gemini 4 Neon 88.1 leads since November 2026
Not in model cards

Language models answer research questions with and without search, and Parallel ranks their accuracy, search lift, cost, and time.

The leaderboard scores 58 models across 100 questions each from DeepSearchQA, Humanity's Last Exam, and Parallel's WISER benchmark. It publishes Search Intelligence, lift, efficiency, time, and cost per correct answer views.

Released
September 2026
Kind
Composite indices
Level
Expert
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text
Interaction
Agentic
Environment
Browser, APIs and tools
Items from
Another benchmark
Refresh
One fixed set
Run by
Anyone
In views
Agents
Strongest model tested
GPT-6.2 Astra #1 on the capability index
Last model added
November 2026
Scores refreshed
November 21st, 2026
58
Entries on its own leaderboard, 51 models we track

Gemini 4 Neon leads with 88.1 Gemini 4 Neon leads with 88.1

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. This view shows only models we track, each at its best.

  1. 1 Gemini 4 Neon 88.1
  2. 2 Claude Fable 5.2 86.9
  3. 3 Claude Opus 5.5 86.7
  4. 4 Gemini 4 Carbon 85.1
  5. 5 GPT-6.2 Astra 84.5
  6. 6 Gemini 4 Argon 84.2
  7. 7 GPT-6.2 Sol 83.7
  8. 8 Gemini 4 Hydrogen 83.0
  9. 9 Tencent Hy4 75.4
  10. 10 Kimi K3.1 Pro 74.2
  11. 11 Claude Sonnet 5.5 74.1
  12. 12 Claude Fable 5.1 73.4
  13. 13 Pareto 26.9 72.5
  14. 14 Muse Wonder 71.3
  15. 15 GPT-6 Astra 70.8
  16. 16 GPT-6.1 Sol 70.4
  17. 17 Claude Opus 5 70.0
  18. 18 Particle-1 69.2
  19. 19 GPT-6.1 Terra 69.2
  20. 20 DeepSeek V4.1 Flash 1017 68.9
  21. 21 MiniMax-M3.1-Flash 68.8
  22. 22 Pareto 26.10 68.6
  23. 23 Ember-1 67.9
  24. 24 OpenRouter Auto (max) 67.7
  25. 25 GPT-5.6 Sol 67.7
  26. 26 Muse Spark Open 67.2
  27. 27 Gemini 3.7 Flash 66.8
  28. 28 Claude Sonnet 5 66.6
  29. 29 GPT-6 Sol 66.6
  30. 30 Muse Spark 1.3 66.1
  31. 31 Qwen 3.8 Max 66.0
  32. 32 Grok 4.7 medium 66.0
  33. 33 Gemini 3.8 Flash 65.1
  34. 34 Tencent Hy4 preview 64.6
  35. 35 MiMo-V2.6 Pro 64.4
  36. 36 Mistral Large 4 64.4
  37. 37 Kimi K3 64.2
  38. 38 Claude Haiku 5.5 62.9
  39. 39 GLM-5.3 62.7
  40. 40 DeepSeek V4.1 Flash 62.7
  41. 41 GPT-6 Luna 61.9
  42. 42 Ultramarine 1 Flash 61.1
  43. 43 Step 5 Preview 61.1
  44. 44 MiMo-V2.6 Flash 60.9
  45. 45 GLM-5.3-Flash 60.8
  46. 46 GPT-5.6 Luna 60.7
  47. 47 Nemotron 3.5 Super VL 59.1
  48. 48 Gemini 3 Flash 58.6
  49. 49 DeepSeek-V4-Pro 58.4
  50. 50 Hunyuan 3 58.2
  51. 51 Beam 56.0
  52. 52 GLM-5.2 54.3
  53. 53 Nemotron 3 Ultra 53.0
  54. 54 MiniMax-M3 52.6
  55. 55 MiMo-V2.5 49.5
  56. 56 Laguna S 2.1 39.5
  57. 57 Nemotron 3.5 Lightning 30B-A3B 37.9
  58. 58 Command A+ 34.8

Source: Official Parallel Search Capability Leaderboard leaderboard

Source: Official Parallel Search Capability Leaderboard leaderboard

#EntryScoreSettingDateSource
1 Gemini 4 Neon 88.1 Nov 2026 Official leaderboard
2 Claude Fable 5.2 86.9 Nov 2026 Official leaderboard
3 Claude Opus 5.5 86.7 Sep 2026 Official leaderboard
4 Gemini 4 Carbon 85.1 Oct 2026 Official leaderboard
5 GPT-6.2 Astra 84.5 Nov 2026 Official leaderboard
6 Gemini 4 Argon 84.2 Sep 2026 Official leaderboard
7 GPT-6.2 Sol 83.7 Nov 2026 Official leaderboard
8 Gemini 4 Hydrogen 83.0 Nov 2026 Official leaderboard
9 Tencent Hy4 75.4 Nov 2026 Official leaderboard
10 Kimi K3.1 Pro 74.2 Oct 2026 Official leaderboard
11 Claude Sonnet 5.5 74.1 Sep 2026 Official leaderboard
12 Claude Fable 5.1 73.4 Sep 2026 Official leaderboard
13 Pareto 26.9 closed weights 72.5 Official leaderboard
14 Muse Wonder 71.3 Oct 2026 Official leaderboard
15 GPT-6 Astra 70.8 Sep 2026 Official leaderboard
16 GPT-6.1 Sol 70.4 Sep 2026 Official leaderboard
17 Claude Opus 5 70.0 Jul 2026 Official leaderboard
18 Particle-1 closed weights 69.2 Official leaderboard
19 GPT-6.1 Terra 69.2 Oct 2026 Official leaderboard
20 DeepSeek V4.1 Flash 1017 68.9 Oct 2026 Official leaderboard
21 MiniMax-M3.1-Flash 68.8 Oct 2026 Official leaderboard
22 Pareto 26.10 closed weights 68.6 Official leaderboard
23 Ember-1 67.9 Sep 2026 Official leaderboard
24 OpenRouter Auto (max) closed weights 67.7 Official leaderboard
25 GPT-5.6 Sol 67.7 Jul 2026 Official leaderboard
26 Muse Spark Open 67.2 Oct 2026 Official leaderboard
27 Gemini 3.7 Flash 66.8 Aug 2026 Official leaderboard
28 Claude Sonnet 5 66.6 Jun 2026 Official leaderboard
29 GPT-6 Sol 66.6 Sep 2026 Official leaderboard
30 Muse Spark 1.3 66.1 Sep 2026 Official leaderboard
31 Qwen 3.8 Max 66.0 Aug 2026 Official leaderboard
32 Grok 4.7 66.0 medium Sep 2026 Official leaderboard
33 Gemini 3.8 Flash 65.1 Sep 2026 Official leaderboard
34 Tencent Hy4 preview 64.6 Aug 2026 Official leaderboard
35 MiMo-V2.6 Pro 64.4 Sep 2026 Official leaderboard
36 Mistral Large 4 64.4 Oct 2026 Official leaderboard
37 Kimi K3 64.2 Jul 2026 Official leaderboard
38 Claude Haiku 5.5 62.9 Oct 2026 Official leaderboard
39 GLM-5.3 62.7 Aug 2026 Official leaderboard
40 DeepSeek V4.1 Flash 62.7 Sep 2026 Official leaderboard
41 GPT-6 Luna 61.9 Sep 2026 Official leaderboard
42 Ultramarine 1 Flash open weights 61.1 Official leaderboard
43 Step 5 Preview 61.1 Sep 2026 Official leaderboard
44 MiMo-V2.6 Flash 60.9 Sep 2026 Official leaderboard
45 GLM-5.3-Flash 60.8 Aug 2026 Official leaderboard
46 GPT-5.6 Luna 60.7 Jul 2026 Official leaderboard
47 Nemotron 3.5 Super VL 59.1 Oct 2026 Official leaderboard
48 Gemini 3 Flash 58.6 Dec 2025 Official leaderboard
49 DeepSeek-V4-Pro 58.4 Apr 2026 Official leaderboard
50 Hunyuan 3 open weights 58.2 Official leaderboard
51 Beam open weights 56.0 Official leaderboard
52 GLM-5.2 54.3 Jun 2026 Official leaderboard
53 Nemotron 3 Ultra 53.0 Jun 2026 Official leaderboard
54 MiniMax-M3 52.6 Jun 2026 Official leaderboard
55 MiMo-V2.5 49.5 Apr 2026 Official leaderboard
56 Laguna S 2.1 39.5 Jul 2026 Official leaderboard
57 Nemotron 3.5 Lightning 30B-A3B 37.9 Aug 2026 Official leaderboard
58 Command A+ 34.8 May 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 4 Neon 88.1 Nov 2026 Official leaderboard
2 Claude Fable 5.2 86.9 Nov 2026 Official leaderboard
3 Claude Opus 5.5 86.7 Sep 2026 Official leaderboard
4 Gemini 4 Carbon 85.1 Oct 2026 Official leaderboard
5 GPT-6.2 Astra 84.5 Nov 2026 Official leaderboard
6 Gemini 4 Argon 84.2 Sep 2026 Official leaderboard
7 GPT-6.2 Sol 83.7 Nov 2026 Official leaderboard
8 Gemini 4 Hydrogen 83.0 Nov 2026 Official leaderboard
9 Tencent Hy4 75.4 Nov 2026 Official leaderboard
10 Kimi K3.1 Pro 74.2 Oct 2026 Official leaderboard
11 Claude Sonnet 5.5 74.1 Sep 2026 Official leaderboard
12 Claude Fable 5.1 73.4 Sep 2026 Official leaderboard
13 Muse Wonder 71.3 Oct 2026 Official leaderboard
14 GPT-6 Astra 70.8 Sep 2026 Official leaderboard
15 GPT-6.1 Sol 70.4 Sep 2026 Official leaderboard
16 Claude Opus 5 70.0 Jul 2026 Official leaderboard
17 GPT-6.1 Terra 69.2 Oct 2026 Official leaderboard
18 DeepSeek V4.1 Flash 1017 68.9 Oct 2026 Official leaderboard
19 MiniMax-M3.1-Flash 68.8 Oct 2026 Official leaderboard
20 Ember-1 67.9 Sep 2026 Official leaderboard
21 GPT-5.6 Sol 67.7 Jul 2026 Official leaderboard
22 Muse Spark Open 67.2 Oct 2026 Official leaderboard
23 Gemini 3.7 Flash 66.8 Aug 2026 Official leaderboard
24 Claude Sonnet 5 66.6 Jun 2026 Official leaderboard
25 GPT-6 Sol 66.6 Sep 2026 Official leaderboard
26 Muse Spark 1.3 66.1 Sep 2026 Official leaderboard
27 Qwen 3.8 Max 66.0 Aug 2026 Official leaderboard
28 Grok 4.7 66.0 medium Sep 2026 Official leaderboard
29 Gemini 3.8 Flash 65.1 Sep 2026 Official leaderboard
30 Tencent Hy4 preview 64.6 Aug 2026 Official leaderboard
31 MiMo-V2.6 Pro 64.4 Sep 2026 Official leaderboard
32 Mistral Large 4 64.4 Oct 2026 Official leaderboard
33 Kimi K3 64.2 Jul 2026 Official leaderboard
34 Claude Haiku 5.5 62.9 Oct 2026 Official leaderboard
35 GLM-5.3 62.7 Aug 2026 Official leaderboard
36 DeepSeek V4.1 Flash 62.7 Sep 2026 Official leaderboard
37 GPT-6 Luna 61.9 Sep 2026 Official leaderboard
38 Step 5 Preview 61.1 Sep 2026 Official leaderboard
39 MiMo-V2.6 Flash 60.9 Sep 2026 Official leaderboard
40 GLM-5.3-Flash 60.8 Aug 2026 Official leaderboard
41 GPT-5.6 Luna 60.7 Jul 2026 Official leaderboard
42 Nemotron 3.5 Super VL 59.1 Oct 2026 Official leaderboard
43 Gemini 3 Flash 58.6 Dec 2025 Official leaderboard
44 DeepSeek-V4-Pro 58.4 Apr 2026 Official leaderboard
45 GLM-5.2 54.3 Jun 2026 Official leaderboard
46 Nemotron 3 Ultra 53.0 Jun 2026 Official leaderboard
47 MiniMax-M3 52.6 Jun 2026 Official leaderboard
48 MiMo-V2.5 49.5 Apr 2026 Official leaderboard
49 Laguna S 2.1 39.5 Jul 2026 Official leaderboard
50 Nemotron 3.5 Lightning 30B-A3B 37.9 Aug 2026 Official leaderboard
51 Command A+ 34.8 May 2026 Official leaderboard

Search Intelligence versus search lift

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 20 40 60 80 100 304050 Lift from search (score) Score with search (score) Gemini 4 Neon 0 20 40 60 80 100 304050 Lift from search (score) Score with search (score) Gemini 4 Neon
ModelSettingScore with searchLift from search
Gemini 4 Neon 88.1 score 45.6 score
Claude Fable 5.2 86.9 score 29.5 score
Claude Opus 5.5 86.7 score 40.2 score
Gemini 4 Carbon 85.1 score 39.5 score
GPT-6.2 Astra 84.5 score 37.4 score
Gemini 4 Argon 84.2 score 45.4 score
GPT-6.2 Sol 83.7 score 35 score
Gemini 4 Hydrogen 83 score 44.5 score
Tencent Hy4 75.4 score 45.3 score
Kimi K3.1 Pro 74.2 score 35.5 score
Claude Sonnet 5.5 74.1 score 40.9 score
Claude Fable 5.1 73.4 score 26.5 score
Pareto 26.9 72.5 score 32.8 score
Muse Wonder 71.3 score 36 score
GPT-6 Astra 70.8 score 27.9 score
GPT-6.1 Sol 70.4 score 30.6 score
Claude Opus 5 70 score 32.8 score
Particle-1 69.2 score 37.8 score
GPT-6.1 Terra 69.2 score 39 score
DeepSeek V4.1 Flash 1017 68.9 score 36 score
MiniMax-M3.1-Flash 68.8 score 41.7 score
Pareto 26.10 68.6 score 33.8 score
Ember-1 67.9 score 38 score
GPT-5.6 Sol 67.7 score 26.3 score
OpenRouter Auto (max) 67.7 score 30 score
Muse Spark Open 67.2 score 31.8 score
Gemini 3.7 Flash 66.8 score 27.4 score
GPT-6 Sol 66.6 score 28.7 score
Claude Sonnet 5 66.6 score 43.8 score
Muse Spark 1.3 66.1 score 35.8 score
Qwen 3.8 Max 66 score 37.1 score
Grok 4.7 medium 66 score 44.8 score
Gemini 3.8 Flash 65.1 score 26.1 score
Tencent Hy4 preview 64.6 score 33.9 score
MiMo-V2.6 Pro 64.4 score 34.7 score
Mistral Large 4 64.4 score 43 score
Kimi K3 64.2 score 33.6 score
Claude Haiku 5.5 62.9 score 39.4 score
DeepSeek V4.1 Flash 62.7 score 35.7 score
GLM-5.3 62.7 score 40.8 score
GPT-6 Luna 61.9 score 33.9 score
Step 5 Preview 61.1 score 33.3 score
Ultramarine 1 Flash 61.1 score 37.1 score
MiMo-V2.6 Flash 60.9 score 38.5 score
GLM-5.3-Flash 60.8 score 40.4 score
GPT-5.6 Luna 60.7 score 33.9 score
Nemotron 3.5 Super VL 59.1 score 42.2 score
Gemini 3 Flash 58.6 score 24.5 score
DeepSeek-V4-Pro 58.4 score 27.9 score
Hunyuan 3 58.2 score 35.6 score
Beam 56 score 27.2 score
GLM-5.2 54.3 score 35.9 score
Nemotron 3 Ultra 53 score 37.7 score
MiniMax-M3 52.6 score 29.5 score
MiMo-V2.5 49.5 score 36.1 score
Laguna S 2.1 39.5 score 27.2 score
Nemotron 3.5 Lightning 30B-A3B 37.9 score 28 score
Command A+ 34.8 score 29.8 score

Search Intelligence versus search lift

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 20 40 60 80 100 304050 Lift from search (score) Score with search (score) Gemini 4 Neon 0 20 40 60 80 100 304050 Lift from search (score) Score with search (score) Gemini 4 Neon
ModelSettingScore with searchLift from search
Gemini 4 Neon 88.1 score 45.6 score
Claude Fable 5.2 86.9 score 29.5 score
Claude Opus 5.5 86.7 score 40.2 score
Gemini 4 Carbon 85.1 score 39.5 score
GPT-6.2 Astra 84.5 score 37.4 score
Gemini 4 Argon 84.2 score 45.4 score
GPT-6.2 Sol 83.7 score 35 score
Gemini 4 Hydrogen 83 score 44.5 score
Tencent Hy4 75.4 score 45.3 score
Kimi K3.1 Pro 74.2 score 35.5 score
Claude Sonnet 5.5 74.1 score 40.9 score
Claude Fable 5.1 73.4 score 26.5 score
Muse Wonder 71.3 score 36 score
GPT-6 Astra 70.8 score 27.9 score
GPT-6.1 Sol 70.4 score 30.6 score
Claude Opus 5 70 score 32.8 score
GPT-6.1 Terra 69.2 score 39 score
DeepSeek V4.1 Flash 1017 68.9 score 36 score
MiniMax-M3.1-Flash 68.8 score 41.7 score
Ember-1 67.9 score 38 score
GPT-5.6 Sol 67.7 score 26.3 score
Muse Spark Open 67.2 score 31.8 score
Gemini 3.7 Flash 66.8 score 27.4 score
GPT-6 Sol 66.6 score 28.7 score
Claude Sonnet 5 66.6 score 43.8 score
Muse Spark 1.3 66.1 score 35.8 score
Qwen 3.8 Max 66 score 37.1 score
Grok 4.7 medium 66 score 44.8 score
Gemini 3.8 Flash 65.1 score 26.1 score
Tencent Hy4 preview 64.6 score 33.9 score
MiMo-V2.6 Pro 64.4 score 34.7 score
Mistral Large 4 64.4 score 43 score
Kimi K3 64.2 score 33.6 score
Claude Haiku 5.5 62.9 score 39.4 score
DeepSeek V4.1 Flash 62.7 score 35.7 score
GLM-5.3 62.7 score 40.8 score
GPT-6 Luna 61.9 score 33.9 score
Step 5 Preview 61.1 score 33.3 score
MiMo-V2.6 Flash 60.9 score 38.5 score
GLM-5.3-Flash 60.8 score 40.4 score
GPT-5.6 Luna 60.7 score 33.9 score
Nemotron 3.5 Super VL 59.1 score 42.2 score
Gemini 3 Flash 58.6 score 24.5 score
DeepSeek-V4-Pro 58.4 score 27.9 score
GLM-5.2 54.3 score 35.9 score
Nemotron 3 Ultra 53 score 37.7 score
MiniMax-M3 52.6 score 29.5 score
MiMo-V2.5 49.5 score 36.1 score
Laguna S 2.1 39.5 score 27.2 score
Nemotron 3.5 Lightning 30B-A3B 37.9 score 28 score
Command A+ 34.8 score 29.8 score

The best score, 88.1, costs $206 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.

NVIDIA Z.ai Google DeepMind Anthropic OpenAI Tencent Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 80.0 100.0 $10 $100 $1,000 $10,000 Cost per task, log scale Gemini 4 Neon GPT-6.2 Sol Tencent Hy4 GPT-6.1 Terra Claude Haiku 5.5 Nemotron 3.5 Lightning 30B-A3B 0.00 20.0 40.0 60.0 80.0 100.0 $10 $100 $1,000 $10,000 Cost per task, log scale Gemini 4 Neon Tencent Hy4

Source: Official Parallel Search Capability Leaderboard leaderboard

The best score, 88.1, costs $206 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.

NVIDIA Z.ai Google DeepMind Anthropic OpenAI Tencent Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 80.0 100.0 $10 $100 $1,000 $10,000 Cost per task, log scale Gemini 4 Neon GPT-6.2 Sol Tencent Hy4 GPT-6.1 Terra Claude Haiku 5.5 Nemotron 3.5 Lightning 30B-A3B 0.00 20.0 40.0 60.0 80.0 100.0 $10 $100 $1,000 $10,000 Cost per task, log scale Gemini 4 Neon Tencent Hy4

Source: Official Parallel Search Capability Leaderboard leaderboard

4 models have held number one on the official leaderboard since its release in September 2026

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

Anthropic Google DeepMind
Own leaderboard 4 leaders

Source: Official Parallel Search Capability Leaderboard 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
20.0 40.0 60.0 80.0 100.0 2026 2027 20.0 40.0 60.0 80.0 100.0 2026 2027

Source: Official Parallel Search Capability Leaderboard 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 Other

All 69 in Other
  • 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. Claude Fable 5.2 64.6
  • 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.
  • Vals Science Index In use Agents are scored on the equal average of BioMysteryBench, MysteryMechanism and Terminal-Bench-Science. GPT-6.2 Astra 68.4%
  • Decision Index Public v0.3 In use Decision models return probabilities over labeled options in public tasks covering knowledge, language, retrieval, tools, and arts. A chance-corrected weighted aggregate combines the task scores. Granite Switch 4.1 30B 51.3
  • 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. Qwen3-Embedding-8B 72.2%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (58 entries).
  • Pass Models scored 51 models we track, plus 7 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Neon.
  • Pass Frontier models 10 of the top 10 have a score, led by GPT-6.2 Astra (#1).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on parallel.ai: 51 rows for models we track and 7 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