How does blind voting work?

When you run a Model Race, the interface hides the model names and replaces them with neutral labels (Model A, Model B, etc.). After you select your preferred response and submit your vote, the labels are revealed. This prevents brand bias from influencing your judgment.

How many credits does a Model Race cost?

Each model in the race costs 1 credit per query. A race between 2 models costs 2 credits; a race between 4 models costs 4 credits. There are no additional charges for the racing interface, leaderboard, or vote recording.

Is my voting data anonymous?

Your individual votes are always private. If you opt in to community rankings, your votes are aggregated anonymously — no personally identifiable information is shared. You can opt out at any time from your account settings.

How is the leaderboard calculated?

Your personal leaderboard uses an Elo-style rating system. Each model's score is updated based on wins and losses relative to the strength of the opponent it beat or lost to. Models that consistently beat strong competition rank higher than those that only beat weaker models.

Feature

Now part of Compare — the leaderboard tab of Vincony's model-comparison super-tool. Open it →

Let the best model win.

Run any prompt against 2+ AI models at once. Responses are shown blind — no model names, no bias. Vote on the best answer and build your personal AI leaderboard based on real usage.

Open in Compare See Plans & Pricing

Find the right model through real-world testing

Benchmark data is theoretical. Your own votes are truth.

Blind Comparison

Model identities are hidden until after you vote. No brand bias, no anchoring — just honest evaluation of which response actually solves your problem best.

Real Usage Data

Your votes are collected from real prompts in real workflows, not synthetic benchmarks. The data reflects which models actually perform best for your specific use cases.

Personal Leaderboard

Every vote builds your personal AI model leaderboard. See at a glance which models win most often for writing, code, research, and more — filtered by your own history.

Community Rankings

Opt in to aggregate your votes with the Vincony community. See how millions of real-world comparisons rank models across categories and use cases.

How it works

Submit your prompt

Enter any prompt and select 2 or more models to race. The same prompt is sent to all selected models simultaneously.

Models race

All models generate their responses in parallel. Responses are displayed side-by-side with model identities hidden behind labels like Model A and Model B.

Vote on the best response (blind)

Read both responses and vote for the one that best answers your prompt. Model identities are revealed after you vote, keeping your judgment unbiased.

What teams use Model Race for

Model evaluation before committing to a specific AI for a long-term project

Team standardization — find which model your whole team agrees produces the best results

Quality benchmarking for specific domains like legal drafting, code generation, or customer support

Cost optimization — identify cheaper models that match the quality of expensive ones for routine tasks

Available on all plans

Model Race is available from Starter ($16.99/mo) and above. Each model in a race costs 1 credit per query — the same as a standard query. Community rankings unlock on Pro ($24.99/mo).

Frequently asked questions

Discover your best AI model today

Open in Compare Compare Lab

Explore more

Compare Lab Consensus Engine Smart Routing Compare All Models Pricing