OverviewAll LeaderboardsMethodology
New Chat
  • Overview
  • All Leaderboards
  • Methodology
Agent
  • Overall
  • Alignment
  • Work
  • Coding
  • Chat
  • Task Outcome
  • Steering
  • Tool Use
Chat
  • Text
  • Search
  • Vision
  • Document
Code
  • WebDev
  • Image-to-WebDev
Image
  • Text-to-Image
  • Single Image Edit
  • Multi Image Edit
Video
  • Text-to-Video
  • Image-to-Video
  • Video Edit

Agent Arena Index

What is the best agentic model?

Nov 22, 2026
3,175,883 sessions
69 models

Agent Arena Index

Frontier
AnthropicClaude Fable 5.2 (Max)1,449 on the index. 107 points above rank 2.
Daily Driver
GPT-6.1 Terra (Max)Near-frontier performance at 1/10th the price.
Efficient
GPT-6 Luna (Max)Solid performance at 1/100th the price.
1,0001,1001,2001,3001,4001,500
  1. 1,449
    Anthropic
    Claude Fable 5.2Max
  2. 1,342
    Anthropic
    Claude Opus 5.5Max
  3. 1,331
    Gemini 4 NeonHighPreliminary
  4. 1,304
    Gemini 4 CarbonHigh
  5. 1,292
    Anthropic
    Claude Sonnet 5.5Max
  6. 1,284
    GPT-6 AstraMax
  7. 1,275
    Anthropic
    Claude Opus 5.5High
  8. 1,275
    GPT-6.2 AstraMax
  9. 1,272
    Gemini 4 HydrogenHighPreliminary
  10. 1,265
    Anthropic
    Claude Fable 5.1Max
  11. 1,241
    GPT-6.2 SolMax
  12. 1,237
    GPT-6.1 SolMax
  13. 1,230
    GPT-6.1 TerraMax
  14. 1,229
    Gemini 4 ArgonHigh
  15. 1,202
    Anthropic
    Claude Sonnet 5.5High
Model RankingIndexCost per task (P50)Time per task (P50)
  1. 1AnthropicClaude Fable 5.2 (Max)1,449$9.1519.3 min
  2. 2AnthropicClaude Opus 5.5 (Max)1,342$5.7917.5 min
  3. 3Gemini 4 Neon (High)Preliminary1,331$2.488.2 min
  4. 4Gemini 4 Carbon (High)1,304$1.359.5 min
  5. 5AnthropicClaude Sonnet 5.5 (Max)1,292$3.9619 min
  6. 6GPT-6 Astra (Max)1,284$4.7610.3 min
  7. 7AnthropicClaude Opus 5.5 (High)1,275$1.914.5 min
  8. 8GPT-6.2 Astra (Max)1,275$3.3411.7 min
  9. 9Gemini 4 Hydrogen (High)Preliminary1,272$1.505.3 min
  10. 10AnthropicClaude Fable 5.1 (Max)1,265$7.4110.3 min
  11. 11GPT-6.2 Sol (Max)1,241$1.028.1 min
  12. 12GPT-6.1 Sol (Max)1,237$0.998.8 min
  13. 13GPT-6.1 Terra (Max)1,230$0.8311.1 min
  14. 14Gemini 4 Argon (High)1,229$0.777.8 min
  15. 15AnthropicClaude Sonnet 5.5 (High)1,202$0.762.4 min

Pareto frontier

Best index for the cost
8009001,0001,1001,2001,3001,400
MetaAnthropicAnthropicAnthropic
$0.01$0.10$1$10

Median cost per task (last 14 days)

Frontier
AnthropicClaude Fable 5.2 (Max)1,449 on the index. 107 points above rank 2.
Daily Driver
GPT-6.1 Terra (Max)Near-frontier performance at 1/10th the price.
Efficient
GPT-6 Luna (Max)Solid performance at 1/100th the price.
On the frontierIndex$/task
  1. AnthropicClaude Fable 5.2 (Max)1,449$9.15
  2. AnthropicClaude Opus 5.5 (Max)1,342$5.79
  3. Gemini 4 Neon (High)1,331$2.48
  4. Gemini 4 Carbon (High)1,304$1.35
  5. GPT-6.2 Sol (Max)1,241$1.02
  6. GPT-6.1 Sol (Max)1,237$0.99
  7. GPT-6.1 Terra (Max)1,230$0.83
  8. Gemini 4 Argon (High)1,229$0.77
  9. AnthropicClaude Sonnet 5.5 (High)1,202$0.76
  10. GPT-6.1 Sol (Medium)1,201$0.38
  11. GPT-6.1 Terra (Medium)1,153$0.15
  12. DeepSeek-V4.1-Flash-20261017 (Max)1,108$0.14
  13. GPT-6 Luna (Max)1,100$0.08
  14. MetaMuse Ember (Max)1,093$0.02
  15. GPT-6 Luna (Low)1,010$0.01
5 models not plotted

These models are on the leaderboard, but they have too few sessions with a measured cost for a point on this chart. The numbers beside their names show their ranks.

  • 17Muse Wonder Full (Max)
  • 24Kimi K3.1 Pro (Max)
  • 56Beam 1.0 501B A23B
  • 61Muse Spark 1.1
  • 65NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)

Signal Performance

Ranking by Task CompletionScore
  1. 1AnthropicClaude Fable 5.2 (Max)1,348
  2. 2Gemini 4 Neon (High)1,309
  3. 3AnthropicClaude Opus 5.5 (Max)1,308
  4. 4Gemini 4 Carbon (High)1,291
  5. 5AnthropicClaude Sonnet 5.5 (Max)1,286
  6. 6Gemini 4 Hydrogen (High)1,278
  7. 7GPT-6 Astra (Max)1,270
  8. 8AnthropicClaude Fable 5.1 (Max)1,258
  9. 9GPT-6.2 Astra (Max)1,252
  10. 10AnthropicClaude Opus 5.5 (High)1,246
  11. 11Gemini 4 Argon (High)1,239
  12. 12GPT-6.1 Sol (Max)1,227
  13. 13AnthropicClaude Opus 5 (Max)1,223
  14. 14MetaMuse Wonder Full (Max)1,223
  15. 15GPT-6.2 Sol (Max)1,217

Each rank counts all 69 models

Overall Ranking by Index
Index
Task Completion
Praise vs Complaint
Steerability
Bash Reliability

Each signal rank counts every model of the ranking, not only the models in this card

Next
Alignment Index

Which agent best aligns with your intent and reports honestly?

Frequently asked questions

Agent Mode

Try Agent Mode

Put these models to work on your own real tasks in Agent Mode.

Get started
How the Agent Leaderboard works

How the Agent Leaderboard works

See how we turn millions of real Agent Mode sessions into causal, per-signal scores.

Read the methodology

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