Venice AI Explained: Private AI, DIEM & VVV Tokenomics

Last updated This date marks a full audit, not a minor edit. Our editing team reviews every claim, figure, and platform detail in line with our editorial guidelines before republishing.
Editorially VerifiedEditorial fact-check process

This article has been reviewed and verified for accuracy by our editorial team. All claims, data points and platform details are cross-referenced against primary sources.

Fact checked by Oluwapelumi Adejumo

  • Data accuracy verified
  • Sources cross-referenced
  • Platform details confirmed
View our fact-checking process
Affiliate DisclosureHow Datawallet is funded

Some links on this page are affiliate links. Datawallet may earn a commission when you sign up through them, at no extra cost to you. Ratings and rankings reflect our own testing and assessment criteria.

Read our full disclosure

Summary: Venice AI gives more than 4 million users access to over 200 text, image, video and audio models through a private, uncensored AI platform. Prompts are not stored on its servers, and conversation history stays in the user's browser.

Its Base-based VVV token connects the platform's business to crypto. On October 1, 2026, annual emissions, the new tokens created each year, fell to 2 million VVV, about 86% below the launch rate. Venice also uses a share of subscription and API revenue to buy and destroy tokens.

Privacy and token supply have limits. Closed models such as Claude and GPT still receive the text of each prompt. Revenue-funded burns remain far smaller than new issuance, while Venice holds more than 30 million of roughly 81 million VVV.

Featured platform

4.9

/5

Our Rating

Our rating is an editorial verdict from hands-on testing of fees, security, liquidity, and features. It is not a paid placement. See our Editorial Methodology for the full framework.
Get Started

Venice AI gives over 4 million users private access to 200+ AI models without storing prompts on its servers. Staking its VVV token unlocks Pro access, yield and DIEM, a token worth $1 of daily AI credit.

Privacy Model

No prompt logging, chats stored on your device

VVV Supply

81.1 million total, 2 million emitted yearly

Pricing

Free tier, paid plans from $18 a month

We may receive a commission when you make a transaction through our links, at no extra cost to you.

What is Venice AI?

Venice AI is a generative AI platform built so that its operator does not retain what users type. Its web and mobile apps generate text, images, video, audio and code. Other software can send requests directly through its API.

ShapeShift founder Erik Voorhees launched Venice in May 2024 as an alternative to ChatGPT. Rather than train a frontier model, the company hosts open-source models on rented graphics processors. It also forwards requests to closed models from OpenAI, Anthropic, Google and xAI on users' behalf.

Venice says it does not log prompts or responses. Chat history stays on the user's device, and the company declines to filter the questions people ask its hosted open-source models. Privacy and open access are central to the product.

Issued on Coinbase's Base network, VVV uses the ERC-20 token standard for Ethereum-compatible networks. Holders can stake VVV, or lock it in a contract, to unlock a paid subscription tier and earn yield. These features place it alongside the tokens covered in our guide to the best AI crypto coins.

When Venice announced its Series A funding round on July 1, 2026, it reported 3.5 million registered users and 1.3 trillion tokens processed each month. Tokens are the chunks of text an AI model reads and writes. The company's About page now cites more than 4 million users.

What is Venice AI?

How Does Venice AI Work?

Venice routes requests between users and a large pool of AI models. The model and payment method determine who can read a prompt and how the request is funded.

1. The Venice Proxy and Local Storage

Every request passes through the Venice proxy. This relay server forwards the prompt to a model over an encrypted connection, then streams the answer back. According to Venice, it retains neither prompt nor response content.

Chat history is encrypted and stored in the browser. Clearing browser data therefore erases it unless a paid user has enabled encrypted backup.

For billing and abuse prevention, Venice still processes operational metadata. Its documentation lists IP addresses, account or wallet identifiers, request timestamps, model selections and usage counts.

1. The Venice Proxy and Local Storage

2. Four Privacy Modes

Each model carries one of four privacy labels that indicate what its operator can see.

Mode
How it works
Who can read the prompt
Anonymous
Venice strips the user's identity
The outside model provider
Private
Model runs under a zero-retention contract
The host, during processing only
TEE
Model runs in a sealed hardware enclave
The enclave, via Venice's proxy
E2EE
Prompt is encrypted on the user's device
Only the verified enclave

A trusted execution environment, or TEE, is a part of a processor that even the machine's owner cannot inspect. Both TEE and E2EE are limited to text models under Venice's TEE and E2EE guide. E2EE also disables web search, file uploads and tool use. NEAR AI has supplied part of this capacity since March 2026.

Every Claude, GPT and Gemini model on Venice's price list is labelled Anonymous. Those providers still receive the prompt text.

3. Model Catalog and Uncensored Models

The catalog contains more than 200 models, including open-source options such as DeepSeek, Qwen and Venice's own Uncensored series. Closed models include Claude Opus 5.5, GPT-6 and Grok.

The "uncensored" description applies to the open-source models. Voorhees told TechCrunch that Venice adds no restrictions to them. Closed models retain their makers' rules. Venice's terms of service also prohibit illegal use and material that sexually exploits children.

3. Model Catalog and Uncensored Models

4. API, Agents and x402 Payments

Venice's API follows OpenAI's format, allowing developers to use existing code by changing one web address and one key. It provides access to the same models as the apps. In July 2026, Venice reported about 2 million API calls a day.

Software agents that act without a person approving each step can generate their own API key from a wallet holding staked VVV. Alternatively, they can pay per request in USDC on Base or Solana using the x402 payment standard. No account is required for those payments.

4. API, Agents and x402 Payments

5. DIEM: Tokenized Daily Compute

Venice's second token, DIEM, provides $1 of Venice credit per day for each token staked. The allowance resets at 00:00 UTC, and unused credit does not carry over.

Minting new DIEM requires locking staked VVV at the Mint Rate, which specifies how many staked VVV are needed for one DIEM. Venice's technical breakdown set the starting rate at 90. The rate rises steeply as supply approaches a target set by the company, which increased from 38,000 to 40,000 DIEM in 2026.

To recover the underlying VVV, a minter must burn the same amount of DIEM. On October 10, 2026, DIEM traded near $1,607 on CoinGecko. At that price, recovering the purchase cost takes about 4.4 years of fully used daily credit.

5. DIEM: Tokenized Daily Compute

How to Use Venice AI

A browser is enough to start chatting on Venice. Using its token features requires a Base wallet and some ETH to cover network fees.

From first prompt to API access:

  1. Open the app: Visit venice.ai or install the iOS or Android app. The free plan includes 10 text prompts and 15 image prompts a day.
  2. Check the privacy label: Read the selected model's privacy mode before sharing sensitive information. Anonymous models send the prompt text to an outside provider.
  3. Choose how to pay: Pay $18 a month for Pro by card or crypto, or stake 100 VVV for the same access.
  4. Fund a Base wallet: Purchase VVV on an exchange or a Base trading app and keep some ETH on Base for fees. Our guides to adding Base to MetaMask and bridging to Base explain the setup.
  5. Stake on the token dashboard: Connect your wallet at venice.ai/token and stake through the dashboard. Venice warns that tokens sent directly to a contract address cannot be recovered.
  6. Mint or buy DIEM: For fixed daily API credit, lock staked VVV at the current Mint Rate or purchase DIEM on the open market.
  7. Stake DIEM for credit: Every staked DIEM adds $1 of daily credit. Spending requires an account balance of at least 0.1 staked DIEM.
  8. Create an API key: Generate a key in the API settings. Set a daily spending cap so one application cannot exhaust the allowance.
  9. Plan the exit: DIEM takes one day to unstake. VVV takes seven days, and the tokens cannot be sold during that period.
How to Use Venice AI

Venice AI Pricing

Venice offers subscriptions, pay-as-you-go credits and token staking. Privacy protections are the same across plans. Model access, usage limits and credits vary by tier.

Plan
Monthly price
Yearly price
Monthly credits
Main limits
Free
$0
$0
0
10 text prompts and 15 image prompts a day
Pro
$18
$180
100
Unlimited text, 1,000 images a day
Pro Plus
$68
$735
7,500
Higher image limits, two-month credit rollover
Max
$200
$2,160
22,500
Highest limits, three-month credit rollover

Video, music, premium models and API calls use Venice's internal credits. At the pricing page's fixed rate of 100 credits to $1, Pro Plus includes $75 of monthly usage and Max includes $225.

The API charges per million tokens at published model rates. Venice Uncensored 1.2 costs $0.20 for input and $0.90 for output, compared with $4.80 and $24.00 for Claude Opus 5.5. Web search costs an additional $10 per 1,000 requests.

With VVV at $21.75 on October 10, the 100-token stake needed for Pro costs about $2,175. That is roughly 12 years of annual Pro fees. Holders retain ownership of the stake and earn yield, though its dollar value fluctuates with the token price.

VVV Tokenomics & Supply

Launched on January 27, 2025, VVV began with 100 million tokens and no presale. There is no fixed maximum supply, and holders receive no voting rights. Venice's launch announcement describes the token as a key to its computing capacity.

Allocation and Supply

An airdrop, a free distribution to eligible wallets, allocated half the launch supply to the public. Venice, its team and two company-controlled pools retained the other half.

  • Airdrop: 50 million VVV, divided equally between more than 100,000 Venice users and Base AI projects such as Virtuals Protocol.
  • Venice: 35 million VVV. This included 10 million for the team, with a quarter unlocked at launch and the remainder released over 24 months.
  • Incentive fund and trading pool: 10 million VVV for grants, plus 5 million to seed the first market on Aerodrome, a Base trading app.

Most of the airdrop went unclaimed. According to The Block, 32.6 million tokens were burned on March 12, 2025. That destruction explains why total supply remained below its starting level at about 81.1 million VVV on October 10, 2026, according to CoinGecko. About 48.3 million circulate, while Venice holds more than 30 million.

VVV's $21.75 price put its market capitalization at $1.05 billion, 37% below the record price of $34.53 reached in September 2026.

Emissions, Burns and Utility

New VVV is created each year and distributed to stakers through emissions. Venice has reduced the annual rate in stages, with the latest cut detailed in an August update to its tokenomics post.

Effective date
Annual emissions
January 27, 2025
14 million VVV
August 20, 2025
10 million VVV
February 2026
6 million VVV
July 1, 2026
3 million VVV
October 1, 2026
2 million VVV

At 2 million tokens a year, emissions equal about 2.5% of total supply, compared with 14% at launch. Each reduction lowers staking yield because emissions are its only source.

Revenue also funds purchases of VVV that are then destroyed. Since April 15, 2026, every new subscription triggers an automatic purchase of $2 to $10. Since July 17, $5 from every $100 of API credits has funded the same process. Total burns stand at 33.7 million VVV, but about 33.6 million came from the 2025 airdrop burn and a 1 million VVV team buyback.

For supply to shrink, annual burns must exceed 2 million VVV. That requires roughly $43.5 million of purchases at the current price. Transactions are published on Venice's burn tracker.

  • Pro access: A stake of 100 VVV unlocks Pro without a subscription fee.
  • Yield: Yearly emissions are shared among stakers in proportion to their stake.
  • DIEM minting: Stakers can lock VVV to mint DIEM while retaining 80% of the yield on the locked tokens.
  • Agent access: Software agents holding staked VVV can create API keys without a human account.
Emissions, Burns and Utility

Most of Venice's published revenue and usage figures come from the company or its founder. They have not been audited.

Key figures as of October 2026:

  • Users: Venice reported 3.5 million registered accounts on July 1, 2026. Its website now cites more than 4 million users.
  • Usage: Monthly processing totals 1.3 trillion tokens, with about 2 million API calls a day and a peak of 300,000 requests an hour.
  • Revenue: TechCrunch reported annualized revenue above $70 million in July. On August 17, Voorhees said on X that it had passed $100 million. Both figures extrapolate a recent period to a full year.
  • Profit: The company says it has been profitable since the first quarter of 2026.
  • Payments: According to Voorhees, about 8% of users pay with crypto.
  • Derivatives: On October 10, CoinGecko showed $232 million in open VVV perpetual futures positions, equal to 22% of market value. These are leveraged contracts with no expiry date. Their trading volume was 7.7 times spot volume.

Leveraged contracts now account for far more VVV exposure traded than direct purchases. According to press reports, US venue Kalshi added VVV to its regulated perpetual futures on September 4, with leverage capped at 1.9 times.

Voorhees attributes most of Venice's growth to closing the feature gap with ChatGPT. Privacy, he says, remains what draws people to try it.

Venice AI vs ChatGPT

Venice was built to challenge ChatGPT. Although Voorhees says the feature gap has nearly closed, the products differ in how they handle data, which models they offer and how customers pay.

Factor
Venice AI
ChatGPT
Models
200+ open-source and closed models from many developers
OpenAI's own models
Conversation history
Saved in the user's browser
Saved to the user's OpenAI account
Prompt logging
None on Venice servers, by company policy
Governed by OpenAI's data policies
Content rules
No added filters on open-source models
OpenAI usage policies and moderation
Payment
Card, crypto, staked VVV or DIEM
Card and app store billing
Token
VVV and DIEM on Base
None

Selecting a GPT model in Venice sends the request to OpenAI without the user's identity attached. OpenAI nevertheless processes the prompt text. For token design, Bittensor, which pays a network of independent model operators, offers a closer comparison.

Venice AI Funding

Voorhees was Venice's sole investor at launch. The business operated for two years without outside money, and its token distribution included no sale to venture funds. In May 2025, Venice opened a $27 million incentive fund to award VVV grants to teams building on its API.

A $65 million Series A followed on July 1, 2026. Dragonfly led the round, joined by Coinbase Ventures, F-Prime, North Island Ventures, Archetype, Liquid2 Ventures and Morgan Creek.

The Block reported the terms of the round as follows:

  • Valuation: The company's shares were valued at $1 billion, with investors receiving an 8.98% stake.
  • Token grant: 1.5 million VVV, subject to a one-year lock followed by even releases over three years.
  • Warrants: Rights to buy another 5 million VVV over eight years for about $66.5 million, under the same lock and release schedule.
  • Use of funds: Venice's first data center, expansion into new markets, acquisitions and hiring.

Selling shares meant the company would not need to sell tokens, Voorhees said. If all warrants are exercised, about 6,000 VVV a day would become available from mid-2027.

Is Venice AI Safe?

Venice removes the most common privacy risk in consumer AI by avoiding a permanent server-side record of conversations. Its terms prohibit both Venice and its model providers from training on user content. With E2EE, users can also check the hardware proof themselves.

Much of that protection still depends on commitments being honored. Private mode relies on zero-retention contracts with Venice and its hosts, while the terms of service reserve Venice's right to monitor information sent through the service. CoinGecko also displays a GoPlus warning that the VVV contract's creator can modify it, including its minting rules.

Risks

The AI service and company-controlled token carry distinct risks. Venice's permissive content rules introduce further exposure.

  • Provider visibility: Anonymous models conceal the user's identity but send the full prompt to OpenAI, Anthropic or Google.
  • Metadata: Billing and abuse prevention involve processing IP addresses, wallet identifiers and usage records.
  • Lost history: Clearing the browser cache or losing a device removes locally stored chats unless backup is enabled.
  • Output accuracy: Venice's terms warn that answers can contain errors and should not replace professional advice.
  • Platform dependence: App stores, payment processors and model suppliers may cut off a service if they reject its content rules.
  • Supply concentration: More than 30 million VVV belong to Venice. Investors can acquire up to another 6.5 million from 2027.
  • Net inflation: Revenue-funded burns remain well below annual emissions of 2 million VVV.2
  • Contract control: Upgraded in August 2025, the VVV contract has no independent governance. Our list of smart contract auditing companies explains what a review covers.
  • DIEM dependence: DIEM represents a claim on Venice's service. The company controls the supply target and could alter how credits work.
  • Regulation: Rules governing staking yield, token classification and unfiltered AI content remain unsettled, including the pending US CLARITY Act.

Venice AI Founders

Venice founder and chief executive Erik Voorhees was an early Bitcoin advocate. He built the betting site Satoshi Dice before founding ShapeShift in 2014. After launching Venice in May 2024, he set out its philosophy in an essay titled "The Separation of Mind and State".

ShapeShift initially allowed customers to trade without identity checks. A later Wall Street Journal investigation alleged that suspect funds had passed through the service. Voorhees brings the same reasoning to AI, describing Venice as a neutral tool that treats users as adults.

Based in Sheridan, Wyoming, Venice operates under terms governed by Wyoming law. Its About page names no wider leadership team, leaving public accountability largely with Voorhees.

Final Thoughts

Venice is among the few crypto projects with a consumer product used by millions and reported profits. For people who do not want their conversations stored, its privacy design addresses a real problem.

Calling VVV "deflationary" requires more scrutiny. Emissions have fallen 86%, yet supply will only shrink when revenue-funded burns exceed 2 million VVV a year. Control of the treasury, DIEM target and contract also rests with one company.

Choose Private, TEE or E2EE models for sensitive work, and back up chats. Buyers should approach VVV as a volatile bet on Venice's revenue and account for the seven-day exit delay.

Datawallet Crypto Newsletter

Get the market brief 100,000+ investors read

3+ issues a week Data-driven & independent Free forever

No spam · Unsubscribe anytime

Portfolio tracker

Welcome back

Sign in to see your portfolio.

At least 8 characters.

By creating an account you agree to our terms and privacy policy.