ENفا
Get API key
MODELSArena’s interaction judge tests what agents actually finish▼ 22%DeepSeek V4.1 Flash API price#1Claude Fable 5.2 · 30.48% win shareFUNDINGAnthropic targets $64.4B Nasdaq IPO as revenue surgesTOOLSCursor tops the tools directory this weekMODELSGoogle cuts Gemini 4 Carbon prices before Gemini 2.5 sunsetPOLICYFive frontier AI labs face coordinated SF safety lawsuits
Vigas Arena · updated just now

The best Vision models,
ranked by real use.

Measures multimodal accuracy across UI screenshots, charts, scanned documents, diagrams, and camera image reasoning.

378,706,203sessions rankedHow we rank →

Rising this week · Vision

Claude Fable 5.2up 2.51 pts · #1 in VisionClaude Opus 4.7up 1.60 pts · #2 in VisionMuse Sparkup 0.98 pts · #5 in Vision

The podium · Vision

Win share · this week
  1. Rank 1: Claude Fable 5.2Anthropic · 1309 score21.10%
  2. Rank 2: Claude Opus 4.7Anthropic · 1300 score19.80%
  3. Rank 3: Qwen 3.8 MaxAlibaba · 1299 score19.55%

Full ranking

Vision leaderboard: win share, score, cost per task and votes
#ModelVotes7 days
1Claude Fable 5.2 (Max)Anthropic21.10%1309 ±7$8.02164,000
2Claude Opus 4.7Anthropic19.80%1300 ±7$4.90152,733
3Qwen 3.8 MaxAlibaba19.55%1299 ±7$0.54144,466
4Gemini 4 Carbon (High)Google19.20%1296 ±6$1.42160,199
5Muse SparkMeta18.60%1294 ±9$0.62131,932
6Gemini 4 Hydrogen (High)Google17.40%1284 ±8$0.92122,665
7GPT-6 Astra (Max)OpenAI16.80%1279 ±8$4.10118,398
8Gemini 4 Argon (High)Google14.10%1258 ±7$0.70108,131
#121.10%
Score 1309 ±7$8.02/task164,000 votes
#219.80%
Score 1300 ±7$4.90/task152,733 votes
#319.55%
Score 1299 ±7$0.54/task144,466 votes
Score 1296 ±6$1.42/task160,199 votes
#518.60%
Score 1294 ±9$0.62/task131,932 votes
Score 1284 ±8$0.92/task122,665 votes
#716.80%
Score 1279 ±8$4.10/task118,398 votes
#814.10%
Score 1258 ±7$0.70/task108,131 votes

Pareto frontier

score vs $/task

Green models give the best score for their price. Tap any dot to inspect value vs. performance.

Claude Fable 5.2Pareto optimal
Score 1309 · 21.10% win · $8.02/task
API pricing

Live sessions

LIVE

No sessions in the last 24 hours yet. Send a request through the Vigas API.

Connect via API — start in 30s
§ Methodology

How the rankings are made.

Real sessions, not benchmarks

Win share counts how often a model finishes real tasks that users accept — across real API workloads, coding agents, and blind Arena tests.

Blind pairwise votes

In Arena, people compare two hidden models and vote. Scores use a Bradley–Terry model with 95% confidence intervals.

Cost is part of the score

Every ranking shows dollars per completed task, so you can see who is best for the money — not just best.

§ Vision · Questions & explore

About the vision board.

Vision & video toolsModels newsAPI pricesAPI quickstart

What is the #1 AI model on the Vision leaderboard right now?

Claude Fable 5.2 (Anthropic) leads vision with a 21.10% real-work win share and a Bradley-Terry score of 1309, ahead of #2 Claude Opus 4.7 (19.80%).

Which top-10 model offers the lowest cost per task for vision?

Qwen 3.8 Max balances #3 performance (score 1299) at $0.54 per completed task on Vigas.

How are Vigas leaderboard scores and win shares calculated?

Vigas combines real session outcomes across API workloads and agent harnesses with blind pairwise Arena votes using a Bradley-Terry statistical model with 95% confidence intervals.

Can I call the top vision models through one API?

Yes. Point any OpenAI-compatible client at https://api.vigas.ai/v1 with your Vigas API key or set model to vigas/auto to route across providers automatically.