Fees converted into AI credits: Orbio hits $90M market cap in 20 days, but demand remains unproven

Original author: KarenZ, Foresight News

A transaction fee generated by a token trade, when it reaches Orbio, may turn into credits consumed by AI when writing code or conducting research.

Under Orbio's design, half of the ORBIO transaction fees are used to support AI inference credits. After participants obtain credits, they can either use them themselves or sell them at a discount to developers who need to call models. Thus, traders contribute fees, credit holders put idle balances to work, and AI users have an opportunity to lower their inference costs.

This mechanism attempts to accomplish an unusual conversion: turning the trading heat of the token market into AI resources that developers can actually use.

The market has already begun pricing in this idea. As of the time of writing on September 21, 2026, GMGN data shows that the ORBIO token has a market cap of approximately $82 million, and its market cap briefly rose to $90 million today. For a project that has only been live for about 20 days, the market's expectations are already quite high.

An AI Credit Market Emerging from Pons

To understand Orbio, one can temporarily set the token aside.

From a product perspective, Orbio is an AI credit market. Developers can purchase discounted credits on the platform and then call different models through a unified interface. The official website at the time listed 446 available models. After purchasing or obtaining credits, users can use their own Orbio API Key to call the services.

Its service chain is also clear: users submit requests to the Orbio gateway, and Orbio uses its managed OpenRouter account to call upstream models. Therefore, Orbio's current business focus is on credit allocation, trading, and usage access, while the underlying model services still rely on OpenRouter and the providers it connects to.

The crypto asset side comes from the Pons launch platform. The ORBIO token was issued on Robinhood Chain through Pons, with its launch market paired against tokenized NVDA.

As of September 21, by market cap, among the graduated projects displayed by Pons that day, Orbio ranked second, behind the platform token PONS.

The developer background of the project can also be traced. On the team side, Orbio's publicly listed contributor is Yash (X account @0x_aster). His personal GitHub profile lists development experience on NFT perpetual contract DEX nftperp, and the official nftperp API documentation (updated 2 years ago) also directly bears this account's name, which corresponds to his previous development work.

On September 18, 2026, Jose (@The0xJose) announced that he would serve as an advisor to Orbio. His X bio lists his roles as former founder of nftperp and co-founder and head of product at Pacifica. In follow-up posts to the announcement, Jose also stated that he had known and worked with Yash since the nftperp period, and that he would continue to support his building efforts this time.

Jose's assessment of Orbio focuses mainly on three directions: on-chain infrastructure that supports Agents in continuously executing tasks with minimal human intervention, inference credits that can be tokenized and traded, and a market mechanism that can simultaneously connect credit suppliers and users. These views represent Jose's judgment at the time he joined the project, and they also explain the problem Orbio hopes to solve: how to enable AI credits to be automatically acquired, circulated, and used by programs.

According to TrustMRR's project profile, Orbio was founded in September 2026 and is listed as a one-person team that is self-funded. At least based on currently available public information, it remains an early-stage project with a very small organizational scale and rapid product iteration.

How Does Orbio Operate?

Orbio's mechanism has already undergone one important change. The early version mainly emphasized holding ORBIO to obtain credits. On September 16, the project officially launched CREDIT, further clarifying the path of "staking ORBIO to obtain CREDIT," and airdropped users' previously accumulated available inference balances in the form of CREDIT to the corresponding wallets.

The two tokens play different roles:

• ORBIO: Users can stake it to participate in obtaining CREDIT rewards.
• CREDIT: Priced on the basis that each token corresponds to $1 of Orbio AI usage credits, it can be transferred, sold, or activated into an API balance. Activation burns the corresponding tokens, and the actual amount credited must also take applicable protocol fees into account.

This means that people who want to use AI cheaply can directly purchase CREDIT without first buying and staking ORBIO. Token participants and AI users are connected through the credit market.

The source of funding for the discounts is the key to this design. According to official documentation, half of the ORBIO transaction fees collected are used to support AI usage credits, stakers receive corresponding CREDIT, and then sell the portion they cannot use up.

CREDIT can be understood as an AI usage voucher for which someone has already borne part of the cost. Sellers are willing to cash out at a discount, buyers want to reduce model invocation expenses, and the two parties are matched through the market.

However, the listed discount does not equal the buyer's final savings rate. The official website shows that when purchasing credits through web retail, the platform charges a 5% service fee on the discounted price. Assuming $100 face value of credits is sold for $90, after adding a $4.5 platform fee, the buyer pays a total of $94.5, ultimately saving 5.5% relative to face value.

On September 19, Orbio announced another arrangement for recycling platform revenue: half of platform revenue is used to buy back and stake ORBIO, and the other half is used to purchase inference credits and mint CREDIT accordingly; the credits obtained by the protocol will also be supplied to the order book at a 20% discount.

This means that the Orbio platform itself will also become a credit supplier. At the same time, an additional connection path has been added between product revenue and ORBIO: after the platform generates revenue, it buys ORBIO, but the purchased tokens are staked rather than burned.

Another feature of CREDIT is that it facilitates automatic use by software and Agents. Agents can purchase, receive, and activate CREDIT through smart contracts to replenish their budget for the next task without waiting for manual checkout. However, what goes on-chain is the holding, trading, and activation of credits. The actual model inference and balance accounting still depend on the Orbio gateway and its model providers.

How Has Orbio Performed?

Orbio has already disclosed a set of data that can be used to observe how the product is operating. As of 15:48 on September 21, 2026, the official analytics page shows that the platform has generated a cumulative inference credit value of approximately $159,500, served approximately 249,500 requests cumulatively, and processed approximately 20.5 billion model tokens, the latter including model input and output. Among these, the generated inference credits reflect credit value and cannot be regarded as platform revenue.

In terms of token participation and credit supply, approximately 354 million ORBIO are currently staked, accounting for 37.26% of the current total token supply. The relatively high staking ratio indicates that many holders are participating in CREDIT distribution; however, this alone cannot prove that the credits have been purchased or consumed by end users.

Demand-side data needs to be further distinguished by methodology. The platform has cumulatively recorded 435 purchases and activations, including both retail purchases and on-chain credit activations.

The recorded combined sales scale displayed on the page is approximately $13,150, but this figure combines cash payments and on-chain activations calculated at face value. The official data explanation explicitly states that this metric is not cash revenue.

Another window for observation is TrustMRR. Its page shows Orbio's cumulative revenue as $9,687, and notes that it is verified through the Whop API.

At the very least, these data indicate that real model invocation and credit circulation activity have appeared on Orbio, and the product is not solely about token trading. However, the project is still in a very early validation stage. To judge whether demand is solid, it is still necessary to observe the number of independent paying buyers, repurchase rates, the actual amount of credits consumed, and the platform's net revenue after deducting relevant costs.

How Is It Different from Venice?

Venice is currently one of the more suitable projects to compare with Orbio, because both are trying to connect AI usage rights with on-chain tokens, but their specific mechanisms are not the same.

According to Venice's current documentation, users can stake VVV to obtain sVVV, then lock sVVV to mint DIEM; by staking 1 DIEM, users can obtain $1 per day of Venice usage credits. Unused credits within an epoch do not roll over.

The difference between the two designs is straightforward: Orbio's CREDIT is more like a transferable, sellable prepaid credit that is consumed upon use; Venice's DIEM is more like an on-chain asset that continuously generates a daily usage allowance. Therefore, although both are attempting to tokenize AI credits, one CREDIT cannot simply be equated with one DIEM.

Venice also has a longer business track record. Its announcement on July 1, 2026 disclosed that the platform had 3.5 million registered users, processed approximately 1.3 trillion model tokens per month, and had about 2 million developer API calls per day. These were official disclosed figures at the time, not real-time data as of September 21.

On the token side, VVV currently has a market cap of $1.635 billion and an FDV of $2.77 billion. The market cap gap between it and ORBIO cannot be directly translated into the latter's upside potential: the two differ in business scale, credit rights, supply structure, and value recapture methods.

For example, Venice previously announced in April a mechanism in which qualifying new subscriptions trigger VVV buyback and burn; Orbio's September 19 announcement described buyback and staking. Burning permanently reduces supply, while staked tokens still exist, and the impact of the two on the token cannot be conflated.

Conclusion

Orbio's attempt is to convert part of the transaction fees generated by token trading into AI credits that can be used and sold. Those who obtain the credits can use them themselves or sell them to developers who need to call models.

Compared with simply attaching an as-yet-unrealized AI concept to a token, this mechanism is easier to verify: whether anyone in the market is willing to buy credits, whether discounts can be sustained over the long term, and whether purchased credits are actually consumed will all directly reflect whether the product has demand.

However, ORBIO's current market pricing has clearly run ahead of its business data. What Orbio needs to prove next is whether users can continuously obtain stable price advantages, sufficient purchasable credits, and a reliable invocation experience.