ZenMux AI has introduced two new features aimed at changing how developers and businesses evaluate artificial intelligence models, while simultaneously launching a temporary pricing initiative designed to remove cost from the equation.
The company announced the rollout of a live usage leaderboard called People’s Choice alongside a new Token Receipts feature that gives users a detailed breakdown of token consumption and spending.
The launch comes as competition across the AI industry continues to intensify, with providers competing on performance, speed and affordability.
People’s Choice Ranks Models by Real Usage
The People’s Choice leaderboard tracks actual user activity across supported models and updates rankings in real time.
Instead of relying on benchmark tests or marketing claims, the leaderboard is intended to reflect how customers use models in real world tasks and workflows.
ZenMux says the approach allows the market to determine which systems provide the greatest value when pricing differences are removed.
Token Receipts Aim to Improve Cost Transparency
The company also introduced Token Receipts, a feature that generates itemized records of token usage for every request.
The feature allows users to review exactly how many tokens were consumed during interactions and how those tokens affected overall costs.
Businesses have increasingly requested this level of transparency as AI spending becomes a larger part of technology budgets.
Temporary Price Cuts Target Model Comparisons
As part of the initiative, ZenMux announced a one month experiment that lowers pricing across more than ten leading AI models to levels comparable with DeepSeek pricing.
According to the company, some models will see discounts of up to 80 percent during the promotional period.
The objective is to reduce price as a deciding factor and encourage users to select models based on performance and usefulness rather than cost alone.
AI Industry Continues Shift Toward Cost Competition
Falling inference costs have become one of the defining trends in the artificial intelligence market over the past year.
Developers and enterprises are increasingly comparing models based on practical results, operating expenses and reliability rather than benchmark scores alone.
ZenMux is attempting to position its platform at the center of that shift by creating a marketplace where usage patterns become a public indicator of customer preference.
Market Response Will Be Closely Watched
Whether real world usage proves to be a better measure of model quality than traditional benchmarks remains an open question.
For the next month, however, ZenMux users will effectively serve as judges in a live experiment that could provide insight into how customers choose AI models when pricing barriers disappear.
The results may offer a rare look at which systems developers and businesses prefer when every option costs roughly the same.