A crowdsourced evaluation leaderboard where users compare two AI answers blind and vote to rank models
Information updated 2026-10-20, source: official site and public reports. Prices and benefits may change. Check the official site before you pay.
Pros
- Cited like an industry standard based on millions of human votes
- Compare and try the latest models for free
- 5 million+ monthly users (reported 2026-01)
Cons
- Controversy over voter preference bias (tone, length)
- Neutrality questioned because major labs are also customers
- Not a comparison of apps or services
These are opinions compiled by Pickx from public sources.
Cautions
The domain redirects from lmarena.ai to arena.ai. Conversations may be used as evaluation data, so do not enter sensitive information
At a glance
A crowdsourced evaluation platform that builds model rankings by having people anonymously compare and vote on two AI answers (formerly LMArena, now arena.ai). It was reported to have raised a Series A at a $1.7B valuation in 2026-01 (5 million+ monthly users, 60 million+ conversations per month) and a $200M Series B ($3.1B valuation) in 2026-10.
Main features
| Feature | Description | Notes |
|---|---|---|
| Blind battles | Two models answer the same question and the user votes | Free |
| Leaderboards by field | Text, vision, documents, search, WebDev, image, video | Based on millions of votes |
| Agent Arena | Evaluates agents on tasks such as coding and document analysis | New in 2026 |
| Alignment Index | Measures whether agents act as humans intend | New in 2026 |
| Pareto Frontier | Compares performance against cost |
Plans and revenue structure
| Audience | Product | Price | Features |
|---|---|---|---|
| Users | Chat/voting/leaderboards | Free | Login optional |
| AI labs/companies | Paid evaluation services | Not disclosed | Customers include OpenAI/Google/xAI (reported), $30M+ annualized in 2025-12 |
Who it suits
- People who want to compare the latest models themselves for free
- People who want to check 'which model is number one right now'
Takeaways for Pickx
- User-participation voting rankings are both content and a data asset → an opportunity for a Korean-specialized version
- A structure that turns free B2C traffic into B2B evaluation and data revenue
- Splitting leaderboards by field creates several 'number one' pieces of content
- To prepare for neutrality controversies: disclose the scoring method and any interests
Notes
- Information checked on: 2026-10-20
- Leaderboard
- Series A report (Pulse 2.0, 2026-01)
- Series B report (The SaaS News, 2026-10)
Good for
6 views
Prices converted at 1 USD = ₩1,329.49 (2026-10-19). Original price shown in brackets. Taxes as listed by each provider.
Similar services
A curated directory that introduces AI tools for marketers inside Openads (오픈애즈), NHN AD's marketer community
NHN ADSouth KoreaFreeKorean: Good
FreeVisit siteA Korean-language AI tool collection site that recommends AI tools by task and compares features and prices
Needs checking (individual maker 'aimoa')South KoreaFreeKorean: Good
FreeVisit siteA Korean AI-specialized media outlet that offers AI news and reports together with an 'AI tool finder', a GPTs collection and a use-case DB
AI Matters (에이아이매터스)South KoreaFreeKorean: Good
FreeVisit siteA benchmark and leaderboard site that independently measures and compares the intelligence, speed and price of AI models and API providers
Artificial AnalysisUnited StatesFree plan availableKorean: Average
$417+Visit siteA Korean version of Product Hunt, a social network where IT service makers post products and talk (many AI products)
disquiet (디스콰이엇)South KoreaFreeKorean: Good
FreeVisit siteA 'directory + education' platform that combines an AI tool directory with YouTube (2,000,000+) and AI courses
FuturepediaUnited StatesFree plan availableKorean: Limited
Check siteVisit site
Comments
Sign in to leave a comment.
No comments yet. Be the first.
Your comments and nickname are public. If you see abuse or advertising, report it here and we will review and remove it.