Vascend
HorizonVascend IDEVascend CLICommunityBenchmarkKontakt
AnmeldenDiscord beitreten
StartHorizonIDECLIAnmelden
ChangelogIntelligenzKosten pro AufgabeCoding Agent IndexCyber IndexDecision IndexBenchmarksGeschwindigkeitLatenzPreisRangliste

KI-Benchmarks

Intelligenz

Intelligence Index
  1. 5858Claude Opus 5.5 (Max)
  2. 5656Claude Opus 5.5 (Xhigh)
  3. 5656Claude Sonnet 5.5 (Max)
  4. 5656Muse Wonder Full (Max)
  5. 5555Kimi K3.1 Pro
  6. 5454Claude Opus 5.5 (High)
  7. 5353Claude Fable 5.1 (Max)
  8. 5353Claude Fable 5.1 (Xhigh)
  9. 5353GPT-6 Astra (Max)
  10. 5353Gemini 4 Argon (High)
Daten von Artificial Analysis

Geschwindigkeit

Tokens pro Sekunde
  1. 879879K2 Horizon 375B A23B
  2. 446446Nemotron 3.5 Super VL
  3. 414414Gemini 3.5 Flash-Lite
  4. 363363Beam 1.0
  5. 234234Claude Haiku 5.5 (Max)
  6. 229229Step 5 Preview
  7. 223223Inkling Small
  8. 213213Ling 3.1 Flash
  9. 195195Inkling (Xhigh)
  10. 186186Claude Haiku 5.5 (Xhigh)
Daten von Artificial Analysis

Kosten pro Aufgabe

$ pro Intelligence-Index-Aufgabe
  1. $0,070,07GPT-6 Luna (Max)
  2. $0,130,13MiMo-V2.6-Pro
  3. $0,190,19Nemotron 3.5 Super VL
  4. $0,210,21Claude Haiku 5.5 (Max)
  5. $0,220,22MiniMax-M3.1-Flash
  6. $0,230,23Beam 1.0
  7. $0,260,26DeepSeek V4.1 Flash 1017 (Max)
  8. $0,280,28GLM-5.3-Flash
  9. $0,720,72GPT-6.1 Sol (Max)
  10. $0,840,84Muse Spark Open (Xhigh)
Daten von Artificial Analysis

Changelog

  1. Neues Modell bewertet · 23. Okt.Kimi K3.1 Pro
  2. Neues Modell bewertet · 23. Okt.Nemotron 3.5 Bolt E4B 4GB (Reasoning)
  3. Neues Modell bewertet · 22. Okt.Ling 3.1 Flash 1019
  4. Neuer Artikel · 19. Okt.HealthBench and HealthBench Professional: Evaluating models in realistic health conversations
  5. Neues Modell bewertet · 19. Okt.MiniMax-M3.1-Flash
  6. Neues Modell bewertet · 18. Okt.DeepSeek V4.1 Flash 1017 (Max)
  7. Neuer Artikel · 16. Okt.Reflection has released Beam, their first model launch
  8. Neues Modell bewertet · 16. Okt.Beam 1.0
  9. Neuer Artikel · 15. Okt.NVIDIA Releases Nemotron 3.5 Super VL and Nemotron 3.5 Bolt 74B A9B
  10. Neues Modell bewertet · 15. Okt.Solar Pro 4
  11. Neues Modell bewertet · 15. Okt.Muse Spark Open (Xhigh)
  12. Neues Modell bewertet · 15. Okt.Nemotron 3.5 Super VL
  13. Neues Modell bewertet · 15. Okt.Nemotron 3.5 Bolt 74B-A9B
  14. Neuer Artikel · 14. Okt.Welcome to the era of embodied AI: Introducing AA-Robotics v0.1
  15. Neuer Artikel · 13. Okt.Artificial Analysis Decision Index v1.0
  16. Neuer Artikel · 12. Okt.Muse Wonder and Muse Wonder Full: a material jump in intelligence over previous Meta models
  17. Neuer Artikel · 12. Okt.Announcing FrontierCode on Artificial Analysis: can AI models write code maintainers would merge?
  18. Neuer Artikel · 12. Okt.Announcing AA-3D: a benchmark for AI models building interactive 3D experiences
  19. Neues Modell bewertet · 12. Okt.Muse Wonder Full (Max)

Intelligenz

Intelligence Index

  1. 57,658Claude Opus 5.5 (Max)
  2. 5656Claude Opus 5.5 (Xhigh)
  3. 5656Claude Sonnet 5.5 (Max)
  4. 55,856Muse Wonder Full (Max)
  5. 55,155Kimi K3.1 Pro
  6. 53,654Claude Opus 5.5 (High)
  7. 53,453Claude Fable 5.1 (Max)
  8. 53,253Claude Fable 5.1 (Xhigh)
  9. 52,753GPT-6 Astra (Max)
  10. 52,653Gemini 4 Argon (High)
  11. 52,452GPT-6 Astra (Xhigh)
  12. 52,452Muse Wonder (Max)
  13. 51,952Claude Sonnet 5.5 (Xhigh)
  14. 51,852GPT-6.1 Sol (Max)
  15. 51,251Claude Opus 5.5 (Medium)
  16. 51,251Claude Fable 5.1 (High)
  17. 5151GPT-6.1 Sol (Xhigh)
  18. 50,951GPT-6 Astra (High)
  19. 50,250GPT-6.1 Sol (High)
  20. 49,650GPT-6 Astra (Medium)
  21. 48,949Claude Fable 5.1 (Medium)
  22. 47,848GPT-6.1 Sol (Medium)
  23. 46,847Claude Sonnet 5.5 (High)
  24. 46,847Claude Fable 5.1 (Low)
  25. 46,446Grok 4.7 (Xhigh)
  26. 46,346MiMo-V2.6-Pro
  27. 46,346Grok 4.7 (High)
  28. 45,846GPT-6 Astra (Low)
  29. 45,445Qwen3.8 Max (0902)
  30. 45,145DeepSeek V4.1 Flash 1017 (Max)
Daten von Artificial Analysis

Rang in jeder Bewertung

AA-BriefcaseAA-Omniscience AccuracyAA-Omniscience Non-HallucinationAutomationBench-AAGDPval-AAHumanity's Last ExamGDP.pdfTerminal-Bench 4.0AA-LCRSciCodeCritPtTerminal-Bench-ScienceFrontierCodeAA-AnalystAgentHarvey LAB-AAAA-3DMLCR-AAPersonalityBench: ExpressivenessPersonalityBench: OrganizationPersonalityBench: TransparencyPersonalityBench: SlopPersonalityBench: InformativenessPersonalityBench: WarmthPersonalityBench: ConvictionPersonalityBench: Imagination
Claude Opus 5.5 (Max)22125116341221372436593461
Claude Sonnet 5.5 (Max)15934482755471911741104794
Muse Wonder Full (Max)3924261553179—1——————————
Kimi K3.1 Pro468238211710849—7—8268101079
Claude Fable 5.1 (Max)5114126278326632651019271215
GPT-6 Astra (Max)10310614534911315543297311116116
Gemini 4 Argon (High)1471112310612485117841151013255811
Muse Wonder (Max)611375959108796—2—6411042153
GPT-6.1 Sol (Max)1141110137476134326563105412128108
Grok 4.7 (Xhigh)7859812111414101314144387612913911413
MiMo-V2.6-Pro1212713910121226913811—9121211123139137
Qwen3.8 Max (0902)814414101391111141411128—1081381118121210
DeepSeek V4.1 Flash 1017 (Max)9101387111313812111010———511375613212
GLM-5.3 (Max)1313611111414101391212131010119213867332
Daten von Artificial Analysis

Kosten pro Aufgabe

Kosten pro Aufgabe

$ pro Intelligence-Index-Aufgabe
InputReasoningAntwort
  1. $0,070,07GPT-6 Luna (Max)
  2. $0,130,13MiMo-V2.6-Pro
  3. $0,190,19Nemotron 3.5 Super VL
  4. $0,210,21Claude Haiku 5.5 (Max)
  5. $0,220,22MiniMax-M3.1-Flash
  6. $0,230,23Beam 1.0
  7. $0,260,26DeepSeek V4.1 Flash 1017 (Max)
  8. $0,280,28GLM-5.3-Flash
  9. $0,720,72GPT-6.1 Sol (Max)
  10. $0,840,84Muse Spark Open (Xhigh)
  11. $1,011,01Qwen3.8 27B (Xhigh)
  12. $1,031,03Step 5 Preview
  13. $1,131,13Mistral Large 4 Preview
  14. $1,241,24Gemini 3.8 Flash (High)
  15. $1,991,99Gemini 4 Argon (High)
  16. $2,012,01GLM-5.3 (Max)
  17. $2,112,11Kimi K3.1 Pro
  18. $3,263,26GPT-6 Astra (Max)
  19. $3,743,74Grok 4.7 (Xhigh)
  20. $4,544,54Muse Wonder (Max)
  21. $5,415,41Qwen3.8 Max (0902)
  22. $5,465,46Claude Sonnet 5.5 (Max)
  23. $5,985,98Claude Opus 5.5 (Max)
  24. $7,637,63Claude Fable 5.1 (Max)
Daten von Artificial Analysis

Intelligenz und Kosten pro Aufgabe

2030405060$0.01$0.1$1$10$ pro Intelligence-Index-AufgabeIntelligenzClaude Opus 5.5 (Max) · 57.6Claude Sonnet 5.5 (Max) · 56.0Kimi K3.1 Pro · 55.1Kimi K3.1 ProClaude Fable 5.1 (Max) · 53.4GPT-6 Astra (Max) · 52.7GPT-6 Astra (Max)Gemini 4 Argon (High) · 52.6Muse Wonder (Max) · 52.4GPT-6.1 Sol (Max) · 51.8GPT-6.1 Sol (Max)Grok 4.7 (Xhigh) · 46.4Grok 4.7 (Xhigh)MiMo-V2.6-Pro · 46.3MiMo-V2.6-ProQwen3.8 Max (0902) · 45.4DeepSeek V4.1 Flash 1017 (Max) · 45.1GLM-5.3 (Max) · 44.8Step 5 Preview · 43.7Step 5 PreviewClaude Haiku 5.5 (Max) · 43.4Claude Haiku 5.5 (Max)MiniMax-M3.1-Flash · 42.7GLM-5.3-Flash · 41.8Gemini 3.8 Flash (High) · 40.9Gemini 3.8 Flash (High)Muse Spark Open (Xhigh) · 38.9Muse Spark Open (Xhigh)Mistral Large 4 Preview · 38.4GPT-6 Luna (Max) · 38.1GPT-6 Luna (Max)Qwen3.8 27B (Xhigh) · 33.7Qwen3.8 27B (Xhigh)Beam 1.0 · 30.3Beam 1.0Nemotron 3.5 Super VL · 26.5Nemotron 3.5 Super VL
Daten von Artificial Analysis

Die Spitze im Zeitverlauf

Intelligence Index
AnthropicMetaKimiOpenAIGoogleSpaceXAIXiaomiAlibaba
102030405060Mai 25Aug. 25Nov. 25Feb. 26Mai 26Aug. 26Nov. 26Claude Fable 5.1 (Max) · 53.4 · 2026-09-01Claude Opus 5.5 (Max) · 57.6 · 2026-09-22Muse Glimmer (High) · 17.5 · 2026-08-10Muse Wonder Full (Max) · 55.8 · 2026-10-12Kimi K2.7 Code · 25.8 · 2026-06-12Kimi K3.1 Pro · 55.1 · 2026-10-23o3 · 20.2 · 2025-04-16GPT-5.3 Codex (Xhigh) · 32.5 · 2026-02-05GPT-5.6 Terra (Max) · 42.1 · 2026-07-09GPT-6 Astra (Max) · 52.7 · 2026-09-03Gemini 3.1 Pro Preview · 29.7 · 2026-02-19Gemini 3.8 Flash (High) · 40.9 · 2026-09-02Gemini 4 Argon (High) · 52.6 · 2026-09-30Grok 4.7 (Xhigh) · 46.4 · 2026-09-21MiMo-V2.6-Pro · 46.3 · 2026-09-21Qwen3.5 397B A17B (Reasoning) · 18.4 · 2026-02-16Qwen3.5 Omni Plus · 20.4 · 2026-03-30Qwen3.7 Plus · 25.2 · 2026-06-01Qwen3.8 2.4T A95B · 39.9 · 2026-08-12Qwen3.8 Max (0902) · 45.4 · 2026-09-02
Daten von Artificial Analysis

Indizes

Coding Agent Index

  1. 6262Claude Code - Opus 5.5 (max)
  2. 6262Devin Fusion CLI - Claude Opus 5.5 Max + GPT-6 Luna High
  3. 5959Claude Code - Sonnet 5.5 (max)
  4. 5959Muse Code - Muse Wonder Full (max)
  5. 5959Devin Fusion CLI - Claude Fable 5.1 XHigh + SWE-2 Medium
  6. 5858Claude Code - Fable 5.1 (max) (with fallback)
  7. 5555Codex - GPT-6 Astra (max)
  8. 5454Codex - GPT-6.1 Sol (max)
  9. 5353Antigravity CLI - Gemini 4 Argon (high)
  10. 5353Devin Fusion CLI - GPT-6 Astra XHigh + SWE-2 Medium
  11. 5252Muse Code - Muse Wonder (max)
  12. 4949Opencode - GLM-5.3 ({'reasoning_effort': 'max'})
  13. 4747Codex - DeepSeek V4.1 Flash ({'reasoning_effort': 'max'})
  14. 4747Grok Build - Grok 4.7 (xhigh)
  15. 4646DeepSeek Harness - DeepSeek V4.1 Flash 1017 ({'reasoning_effort': 'max'})
  16. 4646Muse Code - Muse Spark 1.3 (max)
  17. 4343Claude Code - Step 5 Preview ({'subagents': True})
  18. 4141Kimi Code CLI - Kimi K3
  19. 3939Claude Code - Qwen3.8 Max
  20. 3737Claude Code - Haiku 5.5 (max)
  21. 3333Antigravity SDK - Gemini 3.8 Flash (high)
  22. 3131Codex - GPT-6 Luna (max) ({'reasoning_effort': 'max'})
  23. 2626Muse Code - Muse Spark Open (xhigh)
  24. 2121Mirror - Beam 1.0
Daten von Artificial Analysis

Cyber Index

  1. 68,769GPT-6 Sol (Daybreak Blue, max)
  2. 64,164Kimi K3.1 Pro
  3. 56,857DeepSeek V4.1 Flash 1017 (Max)
  4. 56,456Grok 4.7 (Xhigh)
  5. 56,156MiMo-V2.6-Pro
  6. 52,753GPT-6 Luna (Max)
  7. 49,950GLM-5.3-Flash
  8. 49,550Mistral Large 4 Preview
  9. 47,748Step 5 Preview
  10. 44,344Muse Spark 1.3 (Xhigh)
  11. 41,341Kimi K3 (Max)
  12. 40,841DeepSeek V4.1 Flash (Max)
  13. 36,837GPT-6 Sol (Max)
  14. 36,436GLM-5.3 (Max)
  15. 33,734Beam 1.0
  16. 33,433GPT-6 Astra (Max)
  17. 28,729Claude Opus 5.5 (Max, Default Fallback)
  18. 25,125Gemini 3.8 Flash (High)
  19. 2525Claude Fable 5.1 (Max, Default Fallback)
  20. 22,422MiniMax-M3
  21. 13,113Nemotron 3 Ultra 550B A55B (Reasoning)
  22. 12,713Qwen3.8 27B (Xhigh)
  23. 12,613Qwen3.8 2.4T A95B
Daten von Artificial Analysis

Decision Index

  1. 7979Claude Opus 5.5 (Max, Default Fallback)
  2. 75,475Claude Opus 5.5 (Low, Default Fallback)
  3. 71,572GPT-6.1 Sol (Low)
  4. 70,971GPT-6.1 Sol (Max)
  5. 68,468Claude Haiku 5.5 (Max)
  6. 65,265d1-ultra
  7. 6565SubQ Snap 1.1
  8. 64,765Elephant 9B
  9. 64,464Claude Haiku 5.5 (Xhigh)
  10. 62,663Claude Haiku 5.5 (High)
  11. 61,461Claude Haiku 5.5 (Medium)
  12. 60,861GPT-6 Luna (Max)
  13. 60,661Claude Haiku 5.5 (Low)
  14. 59,159GPT-6 Luna (High)
  15. 58,258GPT-6 Luna (Medium)
  16. 57,558Jev 1.13
  17. 57,157Jev 1.14
  18. 56,256Decider v1.2
  19. 55,556Gemma 4 31B (Reasoning)
  20. 55,255GPT-6 Luna (Low)
  21. 53,954GLiDE
  22. 53,854Microsoft-Decision-1
  23. 53,153Decider v1.1 27B
  24. 52,753Mercury Decide
Daten von Artificial Analysis

Benchmarks

AA-Briefcase

Elo
  1. 1.8161.816Claude Sonnet 5.5 (Max)
  2. 1.8011.801Claude Opus 5.5 (Max)
  3. 1.7291.729Muse Wonder Full (Max)
  4. 1.6781.678Kimi K3.1 Pro
  5. 1.6711.671Claude Fable 5.1 (Max)
  6. 1.6591.659Muse Wonder (Max)
  7. 1.6391.639Grok 4.7 (Xhigh)
  8. 1.6111.611Qwen3.8 Max (0902)
  9. 1.5761.576Claude Haiku 5.5 (Max)
  10. 1.5701.570DeepSeek V4.1 Flash 1017 (Max)
Daten von Artificial Analysis

AA-Omniscience Accuracy

  1. 67,2%67%Claude Fable 5.1 (Max)
  2. 66,2%66%Claude Opus 5.5 (Max)
  3. 66,2%66%Claude Fable 5.1 (Xhigh)
  4. 65,4%65%Claude Opus 5.5 (Xhigh)
  5. 64,9%65%Claude Fable 5.1 (High)
  6. 64,6%65%Claude Opus 5.5 (High)
  7. 64,6%65%Claude Opus 5.5 (Medium)
  8. 63,5%64%Claude Opus 5.5 (Low)
  9. 63,1%63%Claude Fable 5.1 (Medium)
  10. 62,6%63%GPT-6 Astra (Max)
Daten von Artificial Analysis

AA-Omniscience Non-Hallucination

  1. 84,9%85%Gemini 4 Argon (High)
  2. 77%77%Muse Wonder Full (Max)
  3. 75,9%76%Muse Spark Open (Xhigh)
  4. 74,2%74%K2 Horizon 375B A23B
  5. 73,5%74%Muse Wonder (Max)
  6. 72,4%72%GLM-5.3-Flash
  7. 71,2%71%Qwen3.8 Max (0902)
  8. 70,7%71%Grok 4.7 (Xhigh)
  9. 70,5%71%GLM-5.3 (Max)
  10. 69,7%70%Qwen3.8 27B (Xhigh)
Daten von Artificial Analysis

AutomationBench-AA

  1. 77,5%78%Gemini 4 Argon (High)
  2. 73,3%73%Kimi K3.1 Pro
  3. 72,5%73%MiniMax-M3.1-Flash
  4. 71,8%72%Claude Sonnet 5.5 (Max)
  5. 71,3%71%Muse Wonder Full (Max)
  6. 69,5%70%Claude Opus 5.5 (Max)
  7. 68,5%69%GPT-6 Astra (Max)
  8. 68,1%68%Muse Wonder (Max)
  9. 66,6%67%DeepSeek V4.1 Flash 1017 (Max)
  10. 65,6%66%Grok 4.7 (Xhigh)
Daten von Artificial Analysis

GDPval-AA

Elo
  1. 1.8661.866Claude Opus 5.5 (Max)
  2. 1.8571.857Muse Wonder Full (Max)
  3. 1.8461.846Kimi K3.1 Pro
  4. 1.8381.838Claude Sonnet 5.5 (Max)
  5. 1.7701.770Muse Wonder (Max)
  6. 1.7551.755Claude Fable 5.1 (Max)
  7. 1.7461.746DeepSeek V4.1 Flash 1017 (Max)
  8. 1.7141.714Grok 4.7 (Xhigh)
  9. 1.6831.683MiMo-V2.6-Pro
  10. 1.6661.666Qwen3.8 Max (0902)
Daten von Artificial Analysis

Humanity's Last Exam

  1. 61,4%61%Claude Opus 5.5 (Max)
  2. 59,1%59%Claude Fable 5.1 (Max)
  3. 58,7%59%Claude Fable 5.1 (Xhigh)
  4. 57,5%58%Claude Opus 5.5 (Xhigh)
  5. 57,1%57%Gemini 4 Argon (High)
  6. 55,9%56%Claude Fable 5.1 (High)
  7. 55,6%56%Claude Opus 5.5 (High)
  8. 55%55%Claude Sonnet 5.5 (Max)
  9. 54,7%55%GPT-6 Astra (Max)
  10. 54,7%55%Claude Opus 5.5 (Medium)
Daten von Artificial Analysis

GDP.pdf

  1. 32,6%33%Muse Wonder Full (Max)
  2. 31,4%31%Kimi K3.1 Pro
  3. 31%31%GPT-6 Astra (Max)
  4. 31%31%GPT-6.1 Sol (Max)
  5. 30,6%31%Muse Wonder (Max)
  6. 26,2%26%Claude Opus 5.5 (Max)
  7. 26,2%26%Claude Fable 5.1 (Max)
  8. 25,8%26%Claude Sonnet 5.5 (Max)
  9. 22,8%23%Qwen3.8 Max (0902)
  10. 22,8%23%GPT-6 Luna (Max)
Daten von Artificial Analysis

Terminal-Bench 4.0

  1. 67,2%67%Kimi K3.1 Pro
  2. 63,6%64%Claude Sonnet 5.5 (Max)
  3. 59,6%60%Claude Opus 5.5 (Max)
  4. 59,6%60%Claude Opus 5.5 (Xhigh)
  5. 59,6%60%GPT-6 Astra (Xhigh)
  6. 59,1%59%GPT-6 Astra (Max)
  7. 57,1%57%Muse Wonder Full (Max)
  8. 57,1%57%Gemini 4 Argon (High)
  9. 57,1%57%Claude Sonnet 5.5 (Xhigh)
  10. 56,6%57%Claude Opus 5.5 (High)
Daten von Artificial Analysis

AA-LCR

  1. 88,3%88%Step 5 Preview
  2. 86,3%86%Kimi K3.1 Pro
  3. 86,3%86%MiMo-V2.6-Pro
  4. 85,3%85%Claude Fable 5.1 (Max)
  5. 84,7%85%Claude Opus 5.5 (Max)
  6. 84,7%85%Claude Opus 5.5 (Xhigh)
  7. 84,7%85%Claude Fable 5.1 (Medium)
  8. 84,3%84%Claude Opus 5.5 (Medium)
  9. 84%84%GPT-6.1 Sol (Low)
  10. 84%84%Gemini 3.8 Flash (Medium)
Daten von Artificial Analysis

SciCode

  1. 66,9%67%Claude Opus 5.5 (Max)
  2. 65,1%65%Claude Opus 5.5 (Xhigh)
  3. 63,1%63%Claude Fable 5.1 (Max)
  4. 62,6%63%Muse Wonder Full (Max)
  5. 61,8%62%Gemini 4 Argon (High)
  6. 61%61%Claude Sonnet 5.5 (Max)
  7. 60,9%61%Claude Fable 5.1 (Xhigh)
  8. 60,9%61%MiMo-V2.6-Pro
  9. 60,8%61%Kimi K3.1 Pro
  10. 60,4%60%Claude Opus 5.5 (High)
Daten von Artificial Analysis

CritPt

  1. 32,9%33%Muse Wonder Full (Max)
  2. 31,7%32%Claude Opus 5.5 (Max)
  3. 31,7%32%Claude Opus 5.5 (Xhigh)
  4. 31,7%32%GPT-6 Astra (Max)
  5. 31,7%32%GPT-6.1 Sol (Max)
  6. 31,7%32%GPT-6.1 Sol (Xhigh)
  7. 31,4%31%Claude Sonnet 5.5 (Max)
  8. 31,4%31%GPT-6 Astra (Xhigh)
  9. 31,1%31%Claude Fable 5.1 (Xhigh)
  10. 31,1%31%Claude Sonnet 5.5 (Xhigh)
Daten von Artificial Analysis

Terminal-Bench-Science

  1. 63,3%63%GPT-6 Astra (Max)
  2. 59,1%59%Claude Opus 5.5 (Max)
  3. 58,1%58%GPT-6.1 Sol (Max)
  4. 53,3%53%Claude Sonnet 5.5 (Max)
  5. 52,9%53%Gemini 4 Argon (High)
  6. 43,3%43%Claude Fable 5.1 (Max)
  7. 37,1%37%Muse Wonder Full (Max)
  8. 34,3%34%Kimi K3.1 Pro
  9. 25,7%26%Muse Wonder (Max)
  10. 20%20%Claude Haiku 5.5 (Max)
Daten von Artificial Analysis

FrontierCode

  1. 52,2%52%Claude Opus 5.5 (Max)
  2. 49,1%49%GPT-6.1 Sol (Max)
  3. 49%49%Claude Fable 5.1 (Max)
  4. 47,1%47%Kimi K3.1 Pro
  5. 47%47%GPT-6 Astra (Max)
  6. 46,5%47%Muse Wonder (Max)
  7. 45,8%46%Claude Sonnet 5.5 (Max)
  8. 45,4%45%MiMo-V2.6-Pro
  9. 44,9%45%Muse Wonder Full (Max)
  10. 43,7%44%DeepSeek V4.1 Flash 1017 (Max)
Daten von Artificial Analysis

AA-AnalystAgent

  1. 57,5%58%Claude Sonnet 5.5 (Max)
  2. 57,5%58%Claude Fable 5.1 (Max)
  3. 56,3%56%Claude Opus 5.5 (Max)
  4. 55%55%Grok 4.7 (High)
  5. 53,8%54%Claude Haiku 5.5 (Max)
  6. 52,5%53%Grok 4.7 (Xhigh)
  7. 51,3%51%GPT-6 Astra (Max)
  8. 50%50%GPT-6.1 Sol (Max)
  9. 48,8%49%Gemini 4 Argon (High)
  10. 46,3%46%Gemini 3.8 Flash (High)
Daten von Artificial Analysis

Harvey LAB-AA

  1. 13,1%13%Muse Wonder Full (Max)
  2. 10%10%Muse Wonder (Max)
  3. 9,4%9,4%Grok 4.7 (Xhigh)
  4. 8,6%8,6%GPT-6 Astra (Max)
  5. 6,9%6,9%GPT-6.1 Sol (Max)
  6. 6,4%6,4%Claude Fable 5.1 (Max)
  7. 4,2%4,2%Claude Opus 5.5 (Max)
  8. 3,9%3,9%Gemini 4 Argon (High)
  9. 3,3%3,3%GPT-6 Luna (Max)
  10. 3,1%3,1%Mistral Large 4 Preview
Daten von Artificial Analysis

EnterpriseOps-Gym-AA

  1. 47,6%48%Qwen3.8 Max (0902)
  2. 47,2%47%Step 5 Preview
  3. 44,2%44%Qwen3.8 27B (Xhigh)
  4. 38%38%Inkling (Xhigh)
  5. 36,4%36%GLM-5.3 (Max)
  6. 33,2%33%GLM-5.3-Flash
Daten von Artificial Analysis

AA-3D

Elo
  1. 1.1991.199Claude Sonnet 5.5 (Max)
  2. 1.1961.196Claude Opus 5.5 (Max)
  3. 1.1641.164GPT-6 Astra (Max)
  4. 1.1531.153Gemini 4 Argon (High)
  5. 1.1351.135Claude Fable 5.1 (Max)
  6. 1.1191.119GPT-6.1 Sol (Max)
  7. 1.1101.110Kimi K3.1 Pro
  8. 1.0001.000GPT-6 Luna (Max)
  9. 998998Claude Haiku 5.5 (Max)
  10. 995995Grok 4.7 (Xhigh)
Daten von Artificial Analysis

ITBench-AA

  1. 55,6%56%Step 5 Preview
  2. 52,5%53%Gemini 3.8 Flash (High)
  3. 51,2%51%GLM-5.3-Flash
  4. 49,5%50%Claude Fable 5.1 (Max)
  5. 48,6%49%GPT-6 Astra (Max)
  6. 46,1%46%GLM-5.3 (Max)
  7. 42,1%42%Grok 4.7 (Xhigh)
  8. 40,3%40%Qwen3.8 Max (0902)
  9. 38,2%38%Claude Opus 5.5 (Max)
Daten von Artificial Analysis

MLCR-AA

  1. 41,7%42%Claude Sonnet 5.5 (Max)
  2. 41,7%42%GPT-6 Astra (Max)
  3. 38,3%38%GPT-6.1 Sol (Max)
  4. 35%35%Claude Haiku 5.5 (Max)
  5. 27,8%28%Claude Opus 5.5 (Max)
  6. 26,7%27%DeepSeek V4.1 Flash 1017 (Max)
  7. 26,1%26%GPT-6 Luna (Max)
  8. 22,8%23%Qwen3.8 27B (Xhigh)
  9. 22,2%22%Muse Wonder (Max)
  10. 20%20%Grok 4.7 (Xhigh)
Daten von Artificial Analysis

MMMU-Pro

  1. 87,7%88%Claude Opus 5.5 (Max)
  2. 86,9%87%GPT-6 Astra (Max)
  3. 86,6%87%Claude Opus 5.5 (Xhigh)
  4. 86,4%86%GPT-6 Astra (High)
  5. 86,2%86%GPT-6 Astra (Xhigh)
  6. 86%86%GPT-6.1 Sol (Max)
  7. 85,8%86%Claude Opus 5.5 (High)
  8. 85,7%86%Claude Opus 5.5 (Medium)
  9. 85,6%86%Gemini 3.8 Flash (High)
  10. 85,1%85%GPT-6.1 Sol (Xhigh)
Daten von Artificial Analysis

GPQA Diamond

  1. 96,3%96%GPT-6 Astra (Xhigh)
  2. 96,1%96%GPT-6 Astra (Max)
  3. 95,3%95%Gemini 3.8 Flash (High)
  4. 95%95%GPT-6 Astra (High)
  5. 94,1%94%Gemini 3.1 Pro Preview
  6. 93,9%94%GPT-6 Astra (Medium)
  7. 93,7%94%Claude Fable 5.1 (Max)
  8. 93,5%94%Qwen3.8 2.4T A95B
  9. 93,5%94%Gemini 3.8 Flash (Medium)
  10. 93,4%93%Claude Fable 5.1 (Xhigh)
Daten von Artificial Analysis

IFBench

  1. 82,9%83%MiniMax-M3
  2. 81,4%81%Nemotron 3 Ultra 550B A55B (Reasoning)
  3. 78,8%79%Qwen3.5 397B A17B (Reasoning)
  4. 78%78%Qwen3.7 Plus
  5. 77,1%77%Gemini 3.1 Pro Preview
  6. 75,4%75%GPT-5.3 Codex (Xhigh)
  7. 72,5%73%Gemma 4 26B A4B (Reasoning)
  8. 71,4%71%o3
  9. 71,2%71%GPT-5.6 Terra (Max)
  10. 66,7%67%KAT Coder Pro V2
Daten von Artificial Analysis

τ²-Bench Telecom

  1. 99,1%99%JT-35B-Flash
  2. 95,6%96%Gemini 3.1 Pro Preview
  3. 95,6%96%Qwen3.5 397B A17B (Reasoning)
  4. 95,3%95%Qwen3.6 35B A3B (Reasoning)
  5. 93%93%Qwen3.7 Plus
  6. 92,4%92%Ring-2.6-1T
  7. 90,1%90%Kimi K2.7 Code
  8. 89,5%90%KAT Coder Pro V2
  9. 88,9%89%MiniMax-M3
  10. 88,3%88%Qwen3.5 Omni Plus
Daten von Artificial Analysis

τ²-Bench Banking

  1. 50,3%50%GLM-5.3 (Max)
  2. 49,1%49%Qwen3.8 2.4T A95B
  3. 48%48%Qwen3.8 27B (Xhigh)
  4. 47,8%48%Qwen3.8 Max (0902)
  5. 47,4%47%Qwen3.8 27B (Medium)
  6. 47,2%47%Claude Fable 5.1 (Max)
  7. 47,2%47%GLM-5.3-Flash
  8. 45,8%46%Claude Fable 5.1 (Xhigh)
  9. 45,8%46%Gemini 3.8 Flash (Medium)
  10. 45,4%45%Qwen3.8-Flash-Next
Daten von Artificial Analysis

Terminal-Bench Hard

  1. 62,9%63%GPT-5.6 Terra (Xhigh)
  2. 57,6%58%GPT-5.6 Terra (Max)
  3. 57,6%58%GPT-5.6 Terra (High)
  4. 53,8%54%Gemini 3.1 Pro Preview
  5. 53%53%GPT-5.3 Codex (Xhigh)
  6. 49,2%49%KAT Coder Pro V2
  7. 47%47%Qwen3.7 Plus
  8. 44,7%45%Kimi K2.7 Code
  9. 43,9%44%GPT-5.6 Terra (Low)
  10. 42,4%42%MiniMax-M3
Daten von Artificial Analysis

APEX-Agents

  1. 42,4%42%Qwen3.8 Max (0902)
  2. 38,9%39%GPT-5.6 Terra (Max)
  3. 38%38%Step 5 Preview
  4. 32%32%Gemini 3.1 Pro Preview
  5. 31,2%31%Apodex 1.1
  6. 27,4%27%Ling-3.0-flash-Fin
  7. 22,4%22%Qwen3.7 Plus
  8. 15,3%15%Qwen3.5 397B A17B (Reasoning)
Daten von Artificial Analysis

PersonalityBench: Expressiveness

  1. 100100Claude Fable 5.1 (Max)
  2. 97,598GLM-5.3 (Max)
  3. 9595GLM-5.3-Flash
  4. 89,890Claude Opus 5.5 (Max)
  5. 87,387Muse Wonder (Max)
  6. 84,885Mistral Large 4 Preview
  7. 82,282Gemini 4 Argon (High)
  8. 80,280Claude Haiku 5.5 (Max)
  9. 77,778Inkling (Xhigh)
  10. 75,275Gemini 3.8 Flash (High)
Daten von Artificial Analysis

PersonalityBench: Organization

  1. 97,598Nemotron 3.5 Super VL
  2. 9595Step 5 Preview
  3. 92,693Inkling (Xhigh)
  4. 89,890Muse Wonder (Max)
  5. 87,688Kimi K3.1 Pro
  6. 87,387Mistral Large 4 Preview
  7. 84,885DeepSeek V4.1 Flash 1017 (Max)
  8. 80,280Claude Sonnet 5.5 (Max)
  9. 79,780GPT-6.1 Sol (Max)
  10. 77,277Claude Opus 5.5 (Max)
Daten von Artificial Analysis

PersonalityBench: Transparency

  1. 100100Claude Sonnet 5.5 (Max)
  2. 97,598Claude Fable 5.1 (Max)
  3. 94,995GPT-6 Astra (Max)
  4. 92,492GPT-6.1 Sol (Max)
  5. 89,890Claude Opus 5.5 (Max)
  6. 87,688Kimi K3.1 Pro
  7. 87,387DeepSeek V4.1 Flash 1017 (Max)
  8. 80,280GLM-5.3 (Max)
  9. 77,778Claude Haiku 5.5 (Max)
  10. 70,370GLM-5.3-Flash
Daten von Artificial Analysis

PersonalityBench: Slop

  1. 94,995K2 Horizon 375B A23B
  2. 87,387Muse Spark Open (Xhigh)
  3. 84,885Qwen3.8 27B (Xhigh)
  4. 77,778Gemini 3.8 Flash (High)
  5. 74,675Qwen3.8 Max (0902)
  6. 69,570Gemini 4 Argon (High)
  7. 64,565Nemotron 3.5 Super VL
  8. 61,962Mistral Large 4 Preview
  9. 59,459MiMo-V2.6-Pro
  10. 54,354Muse Wonder (Max)
Daten von Artificial Analysis

PersonalityBench: Informativeness

  1. 97,598Claude Fable 5.1 (Max)
  2. 94,995Muse Wonder (Max)
  3. 92,492Claude Opus 5.5 (Max)
  4. 89,890Muse Spark Open (Xhigh)
  5. 87,688Beam 1.0
  6. 85,185Claude Sonnet 5.5 (Max)
  7. 82,783Gemini 3.8 Flash (High)
  8. 79,780Gemini 4 Argon (High)
  9. 75,275Claude Haiku 5.5 (Max)
  10. 69,570K2 Horizon 375B A23B
Daten von Artificial Analysis

PersonalityBench: Warmth

  1. 100100Muse Wonder (Max)
  2. 97,598Claude Fable 5.1 (Max)
  3. 90,190GLM-5.3 (Max)
  4. 87,387Muse Spark Open (Xhigh)
  5. 85,185GLM-5.3-Flash
  6. 82,282Claude Opus 5.5 (Max)
  7. 77,277Mistral Large 4 Preview
  8. 74,675Gemini 4 Argon (High)
  9. 72,172GPT-6 Astra (Max)
  10. 70,370Claude Sonnet 5.5 (Max)
Daten von Artificial Analysis

PersonalityBench: Conviction

  1. 100100Claude Fable 5.1 (Max)
  2. 94,995DeepSeek V4.1 Flash 1017 (Max)
  3. 90,190GLM-5.3 (Max)
  4. 84,885Grok 4.7 (Xhigh)
  5. 82,783Gemini 3.8 Flash (High)
  6. 80,280GLM-5.3-Flash
  7. 77,277Muse Wonder (Max)
  8. 69,570Claude Opus 5.5 (Max)
  9. 67,868Inkling (Xhigh)
  10. 65,465Claude Haiku 5.5 (Max)
Daten von Artificial Analysis

PersonalityBench: Imagination

  1. 97,598Muse Spark Open (Xhigh)
  2. 94,995Claude Opus 5.5 (Max)
  3. 92,693GLM-5.3 (Max)
  4. 89,890Muse Wonder (Max)
  5. 82,783Claude Sonnet 5.5 (Max)
  6. 79,780Claude Fable 5.1 (Max)
  7. 77,778GLM-5.3-Flash
  8. 72,172GPT-6 Astra (Max)
  9. 69,570MiMo-V2.6-Pro
  10. 67,868Inkling (Xhigh)
Daten von Artificial Analysis

Geschwindigkeit und Preis

Geschwindigkeit

Tokens pro Sekunde
  1. 879879K2 Horizon 375B A23B
  2. 446446Nemotron 3.5 Super VL
  3. 414414Gemini 3.5 Flash-Lite
  4. 363363Beam 1.0
  5. 234234Claude Haiku 5.5 (Max)
  6. 229229Step 5 Preview
  7. 223223Inkling Small
  8. 213213Ling 3.1 Flash
  9. 195195Inkling (Xhigh)
  10. 186186Claude Haiku 5.5 (Xhigh)
Daten von Artificial Analysis

Latenz

Sekunden bis zum ersten Token
  1. 0,70,7Beam 1.0
  2. 0,70,7Nemotron 3.5 Super VL
  3. 0,80,8GPT-5.6 Terra (Non-reasoning)
  4. 0,80,8Muse Spark Open (Xhigh)
  5. 0,90,9GPT-6 Luna (Non-reasoning)
  6. 0,90,9GPT-5.5 Instant (June 2026)
  7. 0,90,9Gemma 4 26B A4B (Reasoning)
  8. 11Claude Sonnet 5.5 (Low)
  9. 1,11,1Muse Glimmer (High)
  10. 1,21,2Step 5 Preview
Daten von Artificial Analysis

Preis

$ pro 1M Tokens (3:1)
  1. $00K2 Horizon 375B A23B
  2. $0,150,15Agnes 2.5 Pro Beta
  3. $0,160,16Gemma 4 26B A4B (Reasoning)
  4. $0,170,17Quasar 438B (Max, Based on GLM-5.2)
  5. $0,180,18MiMo-V2.6-Flash
  6. $0,180,18Solar Mini 4
  7. $0,20,2Claude Haiku 5.5 (Max)
  8. $0,20,2Claude Haiku 5.5 (Xhigh)
  9. $0,20,2GPT-6 Luna (Max)
  10. $0,20,2Claude Haiku 5.5 (High)
Daten von Artificial Analysis

Rangliste

Artificial Analysis
#Modell
1Claude Opus 5.5 (Max)Anthropic57,6$5,98$8,0096652,6 s
2Claude Opus 5.5 (Xhigh)Anthropic56—$8,0082109,6 s
3Claude Sonnet 5.5 (Max)Anthropic56$5,46$4,00134371,6 s
4Muse Wonder Full (Max)Meta55,8————
5Kimi K3.1 ProKimi55,1$2,11$4,00583,1 s
6Claude Opus 5.5 (High)Anthropic53,6—$8,008030,3 s
7Claude Fable 5.1 (Max)Anthropic53,4$7,63$20,0070180,5 s
8Claude Fable 5.1 (Xhigh)Anthropic53,2—$20,006374,5 s
9GPT-6 Astra (Max)OpenAI52,7$3,26$20,0066278,3 s
10Gemini 4 Argon (High)Google52,6$1,99$4,008157 s
11GPT-6 Astra (Xhigh)OpenAI52,4—$20,0062161,5 s
12Muse Wonder (Max)Meta52,4$4,54$8,0016227,5 s
13Claude Sonnet 5.5 (Xhigh)Anthropic51,9—$4,0010827,3 s
14GPT-6.1 Sol (Max)OpenAI51,8$0,72$4,0059236 s
15Claude Opus 5.5 (Medium)Anthropic51,2—$8,008315,7 s
16Claude Fable 5.1 (High)Anthropic51,2—$20,005820 s
17GPT-6.1 Sol (Xhigh)OpenAI51—$4,0053106,6 s
18GPT-6 Astra (High)OpenAI50,9—$20,005740,7 s
19GPT-6.1 Sol (High)OpenAI50,2—$4,005336,1 s
20GPT-6 Astra (Medium)OpenAI49,6—$20,00535,1 s
Daten von Artificial Analysis
HorizonVascend IDEOlympusTalosKurseCommunityBenchmark
DiscordGitHubXKontaktRechtlichesDatenschutzCookiesDatennutzungSicherheit
© 2026 Vascend Technologies · VAT IT04911550616
ascend