Venice.aiOwn a share of unrestricted intelligence.
Building on the launch of Venice’s API last November, Venice today releases a novel token-economic design, in which stakers enjoy ongoing access to AI inference via the API at zero marginal cost.
VVV is designed primarily for agents, to reduce economic friction. When staking, Agents now enjoy a persistent inference resource for generative text, image, & code, at scale, without paying per request.
VVV is the first airdrop targeting AI agents themselves: 25% of the genesis supply was allocated to AI community protocol accounts on the Base blockchain (Virtuals, and agents like Luna, aixbt, VaderAI, & others).
Another 25% of supply (VVV) was airdropped to over 100,000 Venice users (thank you for pioneering with us).
Read blog: How to claim and stake VVV
There were no pre-sales of any kind.
Jump to Token & Airdrop details
The Full Story
Venice.ai launched in May 2024 as a private and uncensored alternative to ChatGPT.
In a world of growing AI integration into our daily lives, we believe it is crucial that humans are able to interact with machine intelligence without centralized surveillance and control. Every message you send to ChatGPT (or Gemini, or Anthropic, or Perplexity, or Grok…) is stored forever—to be reviewed, to be classified, to be sold, to be hacked, to be granted to any government that cares to request it.
Venice is different.
Venice is here to prevent an Orwellian dystopia. We are an AI safety company. Our philosophy is described in The Separation of Mind and State.
Since launch, Venice has grown to over 450,000 registered users, plus hundreds of thousands of no-account anons. Over 50,000 users are active daily, making over 15,000 inference requests an hour. You can try it without an account here.
Venice generates revenue with Pro subscriptions, and is owned exclusively by its team.
Venice API and the Rise of Agents
In November 2024, Venice released the beta version of its API for Pro users. It enables the same generative text, image, and code functionality found in the app, but via an API designed for 3rd party apps, developers, and AI agents. Around this same time, this latter group—agents—finally arrived in force.
Blog continues after images
Agents are AI entities (basically bots) with varying degrees of automation and functionality. Some are merely crudely automated scripts that people relabel “AI”. Others are more serious innovations. Like crypto, there is much noise and garbage to parse. And like crypto, we believe the signal underneath is transformative.
Agents consume one very important resource: inference. Inference is the intelligence-stream, the mind, of machines.
Agents today are generally pulling their intelligence-stream from the APIs of OpenAI and others. Three significant problems exist with this:
All conversations, images, and other data is stored forever (surveillance)
The content is controlled and the policies are opaque (censorship)
Inference is paid per request and payment is permissioned (friction)
Venice’s app and API already solve the first two.
In December, Venice was added to the Eliza Framework for AI agents
Addressing the third: consider that if in the 90’s you had to pay for every email you sent, email wouldn’t have taken off so dramatically. It was important to be able to write a message and send it, without worrying about cost (even if the cost per message would be very low). Venice now brings the cost of inference for agents to zero: stake VVV and infer for free.
Agents and other Venice API users can now utilize a crypto token as an access key to a share of total inference capacity. Any staker of the Venice token (VVV) has the ability to consume a pro-rata portion of Venice’s inference capacity via its API, on an ongoing basis.
If you stake 1% of VVV, you get 1% of Venice’s growing API capacity, indefinitely. You do not pay per request.
Consider this new model through the eyes of an AI agent seeking inference:
While staking VVV, the agent earns staking yield as well.
Effectively, this means the marginal cost of inference for the agent becomes negative, assuming neither loss nor gain on the token. Utilizing inference through Venice’s API with VVV can thus result in an agent earning income while it hungrily consumes compute as it produces intelligence.
And unlike ChatGPT et al, this intelligence isn’t being piped into the maw of the CCP and NSA.
And so today, Venice advances unrestricted machine intelligence through its API and the VVV crypto token, to offer a new model for the ongoing consumption of private and uncensored inference for generative text, image, and code.
It’s live and functional
How does VVV actually work?
Importantly, all users of the Venice app (and API) can continue using both as they have been. There is no new requirement to use the token. There is no added friction.
What’s new: any entity (human, app, or agent) that chooses to stake VVV may now draw on the Venice API for inference (text, image, and code) at zero-marginal cost up to their limit per day. The entity does not pay per request.
How much inference can an entity consume in this way? Inference limits are calculated as a pro-rata share of total capacity, divided across all stakers. So if you are staking 1% of the staked VVV, then you may consume 1% of Venice’s API capacity, perpetually. You do not need to spend the VVV, you simply stake it and pull inference when you demand it. The capacity of Venice’s API is measured by the “Venice Compute Unit” or VCU.
What is VCU?
In order to rationalize the resource of inference across various models and types (text vs image, 70b model vs 405b, etc), Venice has created a unified variable called the “Venice Compute Unit.” This is a measure of Venice’s current inference capacity available on its API per day.
The higher the capacity, the more valuable each VVV token, as each represents a share of the total.
If Venice API capacity is 10,000 VCU’s, and an agent has 1% of the staked VVV, then it has the right to consume 100 VCUs (1% of total capacity) each day. And the agent earns additional VVV while staking.
A VCU is essentially shorthand for the unified price of various models, currently representing $0.10 of credit against any model.
Thus in the above example, 1% of the staked VVV would entitle you to $10 of credit every day.
Importantly, as Venice continues building out its infrastructure capacity, the VCU figure will tend to rise. Tomorrow, if the capacity has increased to 20,000 VCU’s, then the same amount of VVV now grants access to $20 of inference credit, every day.
So, when you stake VVV:
Inference is free
Emissions-based yield is paid to you
Thus, the cost for agents and other API users is now below zero. Not for some promotional period, not with some committed subscription of any kind, but at any scale that one stakes VVV. Further, there are no restrictive terms on the API. Agents could theoretically control a share of Venice’s API and resell capacity to other agents on any terms they wish.
Tokenomics
The tokenomics of VVV are relatively simple: there is a limited supply of tokens, and staking these tokens grants stakers an ongoing access to inference at zero-marginal cost.
100 million VVV were created at genesis, half of which airdropped to Venice users and crypto x AI community projects on the Base blockchain.
There was no pre-sale. Venice provides transparent liquidity in the public pool on Aerodrome.
Each year, 14m new VVV are created. Thus the inflation rate starts at 14% and falls over time.
The newly created VVV (emissions) are paid to stakers and to Venice, according to something called the Utilization Rate, which is a measure of demand on Venice’s API. Dynamics of the Utilization Rate are described further here.
Demand for VVV is justified by the utility of staking it to own/control a share of Venice’s AI inference capacity. As Venice inference capacity grows, each VVV represents an increasing quantity of the resource. Inference is the primary resource used by AI agents. Any staker of VVV can use their share of inference at their discretion, or resell/trade it to other entities. The more VVV one stakes, the larger the share of Venice’s API capacity one controls on an ongoing basis. Stakers do not need to spend VVV to get the inference; they get ongoing inference merely by staking.
Stakers are also paid a staking yield, determined by the aforementioned Utilization Rate.
As compute becomes more powerful and less expensive, Venice’s inference capacity will rise, thus VVV is not only shielded from the deflationary effect of computing advancements, but benefits from it (a given stake of VVV will represent an increasingly large portion of compute).
All VVV holders benefit when inference demand for the Venice API grows, making VVV appropriately correlated to the growth of the generative AI industry. This aligns the interests of VVV holders and the Venice company, which is the largest holder. Venice’s VVV treasury is public.
VVV Token Details
100 million TGE supply
14 million emitted annually
No governance
No presale
Audit
Purpose: Stake to obtain free private and uncensored AI inference on Venice API (text/image/code) plus staking yield from emissions.
VVV Airdrop
Tl;dr – You are eligible if you were active on the Venice platform from October 31st onwards and had more than 25 points on December 31st at 23:59, or are part of the listed communities on Base.
Airdrop claim page: venice.ai/claim
50 million (50% of total supply)
25 million VVV airdopped to over 100,000 Venice Users (thank you!)
Eligibility: Active user from Oct 1 with at least 25 points
25 million VVV to Crypto AI Community Protocols
VIRTUALS, AERO, DEGEN, AIXBT, GAME, LUNA, VADER, CLANKER, MOR
Some for @NousResearch when their Psyche token launches <3
Roughly 200 registered Coinbase Agentkit Developers
Snapshot date: Dec 31 2024
Expires: 45 days from TGE (March 13, 2025)
Claim here: venice.ai/claim
VVV TGE Distribution
50 million (50%) airdropped to Venice users and crypto x AI community
35 million (35%) granted to Venice.ai
10m (10%) granted to Team, 25% unlocked upfront and remainder streaming over 24 months.
10 million (10%) Venice Incentive Fund
5 million (5%) liquidity deployment
Please join us in the Discord and learn more in the FAQs.
Thank you for helping us build open AI.
ad intellectum infinitum
-The Venice Team
Back to all posts