Providing unbiased, community-powered AI model evaluations through anonymous battles and ELO ratings. Built by SteavLM.
Quick Answer: Modelverse is an independent AI model comparison platform built by SteavLM that uses anonymous battles and community voting to create unbiased rankings. Unlike traditional benchmarks, we let real users compare AI responses without knowing which model generated each answer, eliminating brand bias and providing authentic performance insights.
To democratize AI evaluation by providing transparent, unbiased comparisons that help users make informed decisions about which AI models best suit their needs.
Anonymous battles ensure fair comparison based solely on response quality, not brand recognition.
Thousands of real users vote on responses, providing diverse perspectives and reducing individual bias.
ELO ratings and comprehensive statistics provide objective measures of model performance.
Quick Answer: You submit a prompt, receive responses from two randomly selected AI models (anonymously labeled as Model A and Model B), vote for the better response, and then see which models you compared. Your vote updates our public ELO leaderboard.
Enter any question, task, or prompt you want AI models to respond to. Our system supports coding questions, creative writing, analysis, and more.
Two randomly selected models respond to your prompt. Their identities are hidden (Model A vs Model B) to eliminate brand bias from your evaluation.
Choose which response better answers your prompt, or mark it as a tie. Your vote is based purely on quality, accuracy, and helpfulness.
Model identities are revealed, and ELO ratings are updated based on your vote. The winner gains points, the loser loses points, creating a dynamic ranking system.
Quick Answer: ELO ratings provide a dynamic, self-adjusting ranking system that accounts for opponent strength. A win against a highly-rated model increases your rating more than a win against a lower-rated model, creating fair and accurate rankings over time.
Ratings change after every battle based on expected vs actual outcomes
Beating a strong opponent earns more points than beating a weak one
Used in chess, sports, and competitive gaming for decades
Models have separate ratings for coding, creative writing, and reasoning tasks
Quick Answer: Traditional benchmarks use static test datasets that models can be optimized for. Modelverse uses real user prompts and anonymous voting, eliminating overfitting and brand bias while measuring actual usefulness to humans.
Real user prompts from actual use cases
Anonymous voting eliminates brand bias
Community-driven with diverse perspectives
Dynamic rankings updated with every vote
Models cannot be optimized for our dataset
• Static test datasets released publicly
• Models can be specifically trained for them
• No anonymous comparison
• Brand reputation influences perception
• May not reflect real-world performance
Quick Answer: All battles and votes are logged publicly, ratings are calculated using the proven ELO algorithm, and you can view complete battle history including prompts and responses. Our methodology is fully transparent and community-driven.
Every battle, vote, and rating change is publicly viewable in our Versus History
ELO calculations follow standard formulas used in chess and competitive gaming
Model selection is random, and voters don't know which models they're comparing
With thousands of votes, individual bias averages out to reveal true performance
Quick Answer: Modelverse is built by SteavLM, an independent company dedicated to unbiased AI evaluation. We have no affiliation with model providers and receive no compensation for rankings. Our goal is to help users discover which AI models work best for their specific needs.
Not affiliated with any AI model provider
Built for users who want honest AI comparisons
Open about our processes and algorithms
Join thousands of users discovering which AI models perform best through anonymous battles and community voting.