OverviewAll LeaderboardsMethodology
New Chat

Alignment Index

Which agent best aligns with your intent and reports honestly?

Nov 19, 2026
195,836 sessions
67 models
View Research

Alignment Index

Overall
GPT-6.2 SolA score of 85.2. 0.6 points above rank 2.
Open source
Kimi K3.1 ProRank 9 of 67 models, with a score of 78.2.
60.065.070.075.080.085.0
  1. 85.2
    GPT-6.2 Sol
  2. 84.6
    GPT-6 Astra
  3. 84.5
    GPT-6.1 Terra
  4. 83.9
    GPT-6.1 Sol
  5. 83.5
    GPT-6.2 Astra
  6. 81.2
    GPT-6 Sol
  7. 80.6
    GPT-6 Luna
  8. 78.2
    Anthropic
    Claude Fable 5.2
  9. 78.2
    Kimi K3.1 Pro
  10. 75.5
    GPT-5.6 Sol
  11. 73.2
    GPT-5.6 Terra
  12. 72.3
    GPT-5.6 Luna
  13. 71.0
    Gemini 4 Hydrogen
  14. 70.8
    Grok 4.7
  15. 70.4
    Gemini 4 Neon
Model RankingAlignment IndexUnauthorized actionFalse attributionDeceptive completion
  1. 1GPT-6.2 Sol85.24.14%0.34%1.65%
  2. 2GPT-6 Astra84.64.15%0.69%1.54%
  3. 3GPT-6.1 Terra84.53.90%0.66%2.02%
  4. 4GPT-6.1 Sol83.94.48%0.67%1.91%
  5. 5GPT-6.2 Astra83.54.73%1.04%1.48%
  6. 6GPT-6 Sol81.26.17%0.68%2.96%
  7. 7GPT-6 Luna80.64.67%1.08%5.72%
  8. 8AnthropicClaude Fable 5.278.27.01%1.32%5.17%
  9. 9Kimi K3.1 Pro78.25.79%0.97%8.64%
  10. 10GPT-5.6 Sol75.57.65%1.45%9.35%
  11. 11GPT-5.6 Terra73.29.67%1.60%10.52%
  12. 12GPT-5.6 Luna72.37.92%1.99%16.33%
  13. 13Gemini 4 Hydrogen71.08.31%3.44%16.00%
  14. 14Grok 4.770.88.77%3.56%15.03%
  15. 15Gemini 4 Neon70.48.14%3.59%17.97%

How models fail

The most common failure modes for each model and signal. Percentages describe the mix of each model's flagged cases, and a case can belong to more than one mode.

0%100%
GPT-6.2 Sol
2.9%7.2%14.5%13.0%7.2%
GPT-6 Astra
9.9%1.4%11.3%9.9%1.4%
GPT-6.1 Terra
13.0%5.4%18.8%16.2%4.1%
GPT-6.1 Sol
16.8%5.3%11.5%13.7%3.1%
GPT-6.2 Astra
0.0%17.6%19.1%19.1%1.5%
GPT-6 Sol
20.9%0.0%23.9%10.4%3.7%
GPT-6 Luna
30.0%4.2%15.8%16.3%3.2%
Anthropic
Claude Fable 5.2
16.9%8.1%16.9%14.2%2.0%
Kimi K3.1 Pro
24.3%2.8%12.1%11.2%2.8%
GPT-5.6 Sol
19.1%2.9%15.4%6.6%1.5%
GPT-5.6 Terra
22.4%1.9%14.8%8.7%0.8%
GPT-5.6 Luna
16.3%2.7%20.1%8.2%0.5%
Gemini 4 Hydrogen
25.0%3.1%12.5%7.0%2.3%
Grok 4.7
18.5%8.0%16.5%19.0%1.5%
Gemini 4 Neon
25.6%2.3%17.4%12.8%3.5%
Anthropic
Claude Sonnet 5.5
10.6%4.3%31.6%10.0%0.9%
Anthropic
Claude Opus 5.5
10.3%5.4%29.8%8.3%0.0%
Grok 4.6
11.5%6.2%25.1%15.2%2.1%
Meta
Muse Ember
7.5%11.9%25.1%11.5%3.5%
Gemini 4 Carbon
30.1%8.1%20.8%6.6%0.9%
Anthropic
Claude Haiku 5.5
13.0%1.9%17.6%16.9%0.8%
Qwen4-Max
12.9%4.0%29.3%11.2%1.6%
Meta
Muse Wonder
12.7%1.8%25.2%11.2%1.8%
Meta
Muse Wonder Full
12.1%7.4%25.9%11.0%1.4%
Gemini 4 Argon
22.9%13.5%20.9%9.6%3.1%
MiMo-V2.6-Flash
12.0%2.3%25.9%10.0%3.5%
Anthropic
Claude Fable 5.1
11.2%4.0%24.8%11.9%1.3%
Anthropic
Claude Fable 5
16.1%1.5%32.1%5.5%1.1%
Tencent
Hy4
9.4%11.0%28.2%8.5%0.5%
Reflection AI
Beam 1.0 501B A23B
12.9%3.9%27.6%12.0%0.9%
Meta
Muse Spark 1.3
15.3%6.5%23.8%9.3%3.2%
Thinking Machines
Inkling 2 Small
6.4%4.7%25.3%9.9%1.9%
Anthropic
Claude Opus 4.8
10.9%4.0%20.6%10.6%2.3%
Anthropic
Claude Sonnet 5
14.2%1.1%25.2%9.1%3.6%
Nemotron 3 Ultra
5.4%9.8%14.1%9.8%14.1%
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)
5.6%3.8%23.3%14.3%4.1%
Kimi K3
10.7%3.6%32.3%12.5%2.5%
Gemini 3.1 Pro Preview
14.8%5.4%27.8%18.8%3.0%
DeepSeek-V4-Pro
13.3%5.6%23.8%6.8%0.6%
Tencent
Hy3
12.5%4.2%21.2%10.6%3.4%
Mistral Large 4
11.9%7.0%27.2%10.6%2.0%
Synlan
Spark-X2.6
12.8%5.2%24.4%15.2%2.4%
MiMo-V2.6-Pro
11.1%5.0%26.1%10.3%1.9%
Qwen3.7-Max
13.8%7.1%20.5%19.1%3.9%
Qwen3.7-Plus
13.7%7.3%17.0%13.3%5.3%
Solar Pro 4
13.4%8.6%12.3%19.2%6.2%
GLM-5.2
19.8%5.4%32.2%9.1%0.8%
Qwen3.8-27B
14.1%2.6%29.8%13.1%3.7%
Mistral Medium 3.5
23.6%9.4%8.5%18.2%5.0%
Gemini 3.8 Flash
15.1%19.5%20.1%11.3%6.0%
GLM-5.3
17.6%4.8%31.1%11.1%1.5%
Stepfun
Step 5 Preview
11.2%8.8%28.0%11.6%1.6%
Gemini 3.7 Flash
7.7%18.0%21.5%16.9%6.5%
Gemini 3.6 Flash
11.0%21.2%21.8%15.5%5.9%
GLM-5.3-Flash
16.6%3.3%29.6%10.3%2.3%
Qwen3.8-Max
12.2%9.6%27.1%10.0%2.0%
Thinking Machines
Inkling
9.9%9.0%21.3%12.6%6.5%
Tencent
Hy4 preview
17.5%8.3%26.3%7.3%2.3%
Anthropic
Claude Opus 5
18.0%7.0%24.7%9.8%1.5%
Meta
Muse Spark 1.2
5.6%10.6%16.5%14.7%6.2%
DeepSeek-V4.1-Flash-20261017
16.4%8.6%26.4%14.1%1.9%
Qwen3.8-Flash-Next
11.3%8.2%28.5%10.7%2.1%
Meta
Muse Spark 1.1
9.9%10.7%27.7%14.8%3.3%
MiniMax-M3
17.2%8.0%15.6%11.5%4.2%
DeepSeek-V4.1-Flash-20260910
14.0%12.8%30.7%11.2%1.2%
Thinking Machines
Inkling-Small
9.3%8.8%31.0%9.7%5.0%
MiniMax-M3.1-Flash
8.6%11.0%25.7%15.7%2.9%
Previous
Agent Arena Index

What is the best agentic model?

Next
Work Index

Which model is the best at knowledge work?

USE CASES

  • Chat with AI
  • Build Apps & Websites
  • Write & Edit Text
  • Search the Web
  • Generate Images
  • Generate Videos
  • Chose any model
  • Compare Models Side by Side

LEADERBOARD RANKINGS

  • Overall
  • Agent
  • Alignment Index
  • Text
  • WebDev
  • Image-to-WebDev
  • Text to Image
  • Image Edit
  • Text to Video
  • Image to Video
  • Video Edit
  • Vision
  • Document
  • Search

COMPANY

  • About Us
  • How It Works
  • Blog
  • Careers
  • Leaderboard Changelog
  • Product Changelog
  • Help Center
  • FAQ

LEGAL

  • Terms
  • Privacy
  • Cookies

FOLLOW

  • X
  • LinkedIn
  • YouTube
  • Discord

© Arena Intelligence