Venice

What it does

Venice is a private AI inferenceInferenceRunning a trained AI model to produce an answer. Inference is what happens when you type a prompt into ChatGPT and get a response. The model takes your input, computes a best guess, and returns it.Like asking an expert for their opinion. The training was the decades they spent becoming an expert. The inference is the 30 seconds it takes them to answer your specific question.Read more → platform. You send prompts, Venice routes them through a centralised proxy to distributed GPUGPUGraphics Processing Unit. Originally designed to render video game graphics, GPUs turned out to be exceptionally good at the massively parallel math that AI models need. Modern AI training and inference runs almost entirely on GPUs.Like a factory with 10,000 workers doing the same simple task in parallel, versus a CPU which is more like 10 workers each doing different complex tasks. AI training involves doing simple math a million times per second on a million numbers, which is exactly what the GPU factory is designed for.Read more → providers, and the response comes back without your prompts or outputs being stored anywhere. No logs. No conversation history on their servers. Everything stays in your browser’s local storage.

As of March 2026, Venice offers four privacy modes. Anonymous proxies requests to frontier models (GPT, Claude, Gemini, Grok) while hiding your identity from the provider. Private (the default) runs inference on zero-data-retention infrastructure. TEETEETrusted Execution Environment. A hardware-secured region of a CPU or GPU where code runs in isolation, so even the machine's operator can't read what's happening inside. TEEs give decentralised AI inference privacy guarantees.Like a bank vault inside a bank. The bank owns the building, staffs the lobby, and runs the security cameras. But what's inside the vault is invisible to everyone, including the bank staff, unless the customer opens it.Read more → runs inference inside hardware-isolated secure enclaves operated by NEAR AI Cloud and Phala Network, with remote attestationAttestationA cryptographic proof that a piece of code is running on a specific hardware enclave in an unmodified state. Attestation lets remote users verify that a service is genuinely running what it claims to be running.Like a tamper-evident seal on a medicine bottle. The seal itself doesn't make the medicine safe, but it gives you a way to verify that nobody opened the bottle and swapped the contents before you bought it.Read more → you can independently verify. E2EE encrypts your promptPromptThe text you give an AI model to tell it what to generate. A prompt can be a simple question, a long instruction, a chunk of context plus a task, or a conversation history the model uses to produce its response.Like a brief you give to a junior designer. A vague brief gets a vague result. A detailed brief with context, constraints, and examples gets something usable. The quality of the output depends heavily on the quality of the brief.Read more → on your device before it leaves your browser; it stays encrypted through Venice’s proxy and is only decrypted inside a verified TEE enclave. Neither Venice nor the GPU provider can see your data at any point. TEE and E2EE are Pro features.

The base architecture works like this: your request hits a Venice-controlled proxy server that strips identifying information, then forwards to GPU providers sourced from Akash Network, Hyperbolic, Prime Intellect, NEAR AI Cloud, and Phala Network. In Private mode, providers see individual prompts but not who sent them. In TEE mode, prompts are processed inside hardware enclaves. In E2EE mode, prompts are encrypted before they leave your device. Responses stream back through the proxy, all data purged immediately.

Founded by Erik Voorhees, the serial crypto entrepreneur behind SatoshiDice and ShapeShift. Venice launched in May 2024 with a team of roughly six people. The operation runs out of a Wyoming shell registration, with Voorhees based in Panama. Teana Baker-Taylor, formerly VP of Policy at Circle and UK Director at Binance, is co-founder and COO.

This isn’t a protocol. It’s a company that runs a proxy server. The distinction matters for everything that follows.

Why Venice AI is uncensored, and what that actually means

“Uncensored AI” is a loaded term, so let me be specific about what Venice does and doesn’t do.

Venice serves open-weight models (DeepSeek, Llama, Qwen, Mistral and others) without corporate content policies layered on top. When you send a prompt to ChatGPT, OpenAI’s safety layer decides whether to answer. When you send the same prompt to Claude, Anthropic’s content policy makes the call. These are editorial decisions by private companies about what their AI is allowed to discuss.

Venice removes that layer. The models respond based on their trainingTrainingThe one-time process of teaching a neural network to perform a task by showing it massive amounts of example data and adjusting its internal weights until the outputs are good. Training builds the model; inference uses it.Like the years an apprentice spends learning a trade. You don't see any of the actual work, just thousands of repeated mistakes gradually becoming competence. By the end, the apprentice can do the job. The training was invisible, but the skill is now permanent.Read more →, not a corporation’s content policy. Pro subscribers can disable Safe Mode entirely, which removes even the modelModelA trained neural network that takes inputs (text, images, audio) and produces outputs (more text, classifications, generated content). In DeAI the model is the thing that actually does the work.Like a very experienced apprentice who has spent years watching thousands of masters make furniture. They can't explain how they know when a joint is right, but they can make a chair that looks and functions like a Chippendale. The training is invisible. The output is what matters.Read more →’s built-in guardrails where the model architecture permits it.

In practice, this means Venice handles prompts that centralised providers refuse. Security researchers can discuss vulnerabilities without triggering refusal responses. Creative writers are not blocked from mature themes. Medical researchers get direct answers about drug interactions rather than “please consult a healthcare professional.” I have had Claude refuse to help with completely legitimate research questions about smart contractSmart ContractA program stored on a blockchain that runs automatically when its conditions are met. Smart contracts are how blockchains do anything beyond just transferring tokens — DeFi, NFTs, DAOs, and DeAI infrastructure all run on smart contracts.Like a vending machine. You put in the right input and it produces the expected output, no human operator required. The rules are fixed in the machine itself, anyone can use it, and nobody can stop a transaction in the middle.Read more → exploits. Venice handles these without friction.

What Venice doesn’t do is serve illegal content. CSAM, direct instructions for mass violence, and other clearly illegal material are blocked at the platform level. “Uncensored” means no corporate content policy, not no rules at all. The distinction matters because critics conflate the two.

The real question is whether uncensored inference is a feature or a risk. For Venice, it is both. Security publications have flagged Venice as a tool for generating malware. If a credible real-world crime is traced to Venice-generated content, the regulatory response could be severe. The UK free tier is already blocked under FSMA. Voorhees is betting that the market for uncensored AI is large enough and legitimate enough to outweigh the regulatory risk. So far, that bet appears to be paying off. Venice claims 1.3 million registered users, which suggests real demand if accurate.

Private AI inference: what Venice actually protects

How Private mode works

When you send a prompt to Venice, it hits a centralised proxy server that strips identifying information: your IP address, session data, anything that ties the prompt to you personally. The raw prompt then routes to GPU providers sourced from Akash Network, Hyperbolic and Prime Intellect. Those providers see individual prompts but not who sent them or the broader conversation context. Responses stream back through the proxy, and all data is purged immediately. Nothing is stored server-side.

Compare this with centralised alternatives. OpenAI retains conversations and trains on them by default (you can opt out, but the default is collection). Anthropic retains conversations for safety research and model improvement. Google processes prompts through infrastructure that feeds into its broader advertising and data ecosystem. With Venice, there’s no conversation history on their servers. Everything stays in your browser’s local storage.

For the default Private mode, this is trust-based privacy, not cryptographically provable. The proxy is closed-source. No independent privacy audit has been completed despite being the third-most-requested feature from users (334 votes on Featurebase). Venice claims zero-knowledge inference, but until the architecture is independently verified, users are trusting Voorhees and a roughly six-person team to honour that promise.

TEE and E2EE: cryptographic guarantees

In March 2026, Venice launched TEE and E2EE modes that move beyond trust. TEE mode runs inference inside hardware-isolated enclaves operated by NEAR AI Cloud and Phala Network. Each response includes a verification icon linking to a full attestation report, a cryptographic certificate proving the prompt was processed inside a genuine secure enclaveEnclaveAn isolated region of CPU or GPU memory protected by hardware. Code and data inside the enclave are inaccessible to the operating system, the hypervisor, or even the machine's physical owner.Like a secure room inside a much larger office building. The building's caretakers have keys to every other room but not this one. What happens inside is invisible to them by design.Read more →. You can independently verify this without taking Venice’s word for it.

E2EE mode goes further. Your prompt is encrypted on your device before it leaves your browser. It stays encrypted through Venice’s proxy. It’s only decrypted inside a verified TEE enclave. Neither Venice nor the GPU provider can see your data at any point. The trade-off is functionality: E2EE disables web search and memory, and responses are slower. TEE mode retains full features including file uploads and auto-routing.

Supported TEE/E2EE models include Venice Uncensored 1.1, GLM 5, Qwen 3.5 122B, Gemma 3 27B, and GPT OSS variants, with more being added. Both modes are Pro features. TEE and E2EE are available via APIAPIApplication Programming Interface. A structured way for one piece of software to talk to another. In DeAI, APIs let applications request inference from a model without running the model themselves.Like a waiter in a restaurant. You don't walk into the kitchen and cook your own meal. You tell the waiter what you want, they tell the kitchen, the kitchen cooks it, and the waiter brings it back. The API is the waiter.Read more → as well as the web interface: developers select models by prefix (tee- or e2ee-) and E2EE requires client-side ECDH key exchange (secp256k1) with AES-256-GCM encryption, three custom headers, and streaming mode. Full implementation docs at docs.venice.ai.

Who needs which mode

For many use cases, the default Private mode is sufficient. If you are a journalist, lawyer, healthcare professional, or simply someone who doesn’t want a corporation indexing your AI interactions, Venice delivers materially more privacy than any mainstream alternative. And if you need cryptographic guarantees, Venice now offers them through TEE and E2EE, powered by the same infrastructure projects (Phala, NEAR) that we’ve reviewed as standalone investments.

Who benefits most: anyone in regulated industries handling sensitive data, anyone in jurisdictions with speech restrictions, developers building products that handle user data, and anyone who considers their AI interactions to be private by default. For a deeper look at why privacy matters for AI specifically, see Privacy Is the Killer App.

Value proposition

Venice sells privacy without friction. You open a browser, type a prompt, and get an uncensored response from open-source models (DeepSeek, Llama, Qwen, Mistral, FLUX and others) without creating an account, providing an email, or agreeing to content restrictions. Pro subscribers can disable Safe Mode entirely. The free tier gives you 10 text prompts and 15 image generations per day.

The API is OpenAI-compatible. If you have existing code calling OpenAI, you change one base URL and your prompts now route through Venice’s privacy proxy instead. For developers, this is the killer feature. I use it daily.

The tokenTokenA digital unit of value or access rights tracked on a blockchain. Tokens can represent ownership in a project, a right to use a service, a share of future revenue, or simply a tradable asset with no underlying claim.Like a physical poker chip a casino issues. The chip itself has no value. What makes it worth something is what it lets you do at the casino, what the casino has promised, and how much other people will pay you for it.Read more → mechanics add a DeFiDeFiDecentralised Finance. Financial services like lending, trading, and yield farming built on smart contracts instead of traditional banks or brokerages. DeFi protocols are usually permissionless and global.Like a vending machine that can give you a loan, swap your currencies, or invest your savings. Nobody is behind the counter, the rules are written into the machine itself, and anyone with money in the right format can use it.Read more → angle. Stake VVV to earn emissions-based yield. To access inference via tokens, you lock staked VVV (sVVV) to mint DIEM. Each DIEM represents $1/day of perpetual, renewing API credit against any model on Venice (including Claude, GPT-5, Gemini and all open-source models). The mint rate follows an exponential curve: as DIEM supply approaches the target, the lock requirement rises sharply (more sVVV must be posted as escrow per DIEM minted). Burning DIEM unlocks your original sVVV. While locked, you still earn 80% of standard stakingStakingLocking up a cryptocurrency to help secure a blockchain network, usually in exchange for rewards. The locked tokens act as a security deposit that can be taken away if the staker misbehaves.Like putting down a large rental deposit for an apartment. You get the money back if you behave, you earn interest while it's locked, and the landlord takes it if you trash the place.Read more → yields. Over 8.26 million VVV is currently held in escrow for DIEM, about 10% of effective supply. DIEM supply has now passed the 38,000 target (38,182 outstanding as of 23 April 2026), pushing the lock requirement into the steeper part of the exponential curve. We track the live mint rate trajectory from on-chain data at /projects/venice/diem/. Minimum threshold is 0.1 DIEM to activate API credits.

A significant capacity optimisation changed DIEM allocation from dividing across all stakers to dividing among active API users only (those who made an API call in the last seven days). The result: a 14x increase in DIEM capacity per token for actual users. This rewards genuine usage rather than passive staking.

The competitive position is clear. If you want private AI inference that actually works today, Venice is the only consumer product delivering it at scale. Venice reports 1.3 million registered users and 45 billion LLMLLMLarge Language Model. A neural network trained on vast amounts of text to predict the next word in a sequence. Modern LLMs (GPT, Claude, Llama, Qwen, DeepSeek) generate human-quality text and are the foundation of most modern AI products.Like an autocomplete that read every book ever written. It has no memory of individual texts but it has absorbed the patterns of language so deeply that it can generate paragraphs that sound human. The skill is statistical, not conscious.Read more → tokens processed daily. Warden Protocol alone claims to drive roughly 1.5 million daily chats through the Venice backend.

The product ships fast. Nine modalities launched within 18 months:

• Text and code
• Image, video, music, audio
• Web search
• Voice mode
• Memoria (local vector memory)

Tokenomics

VVV launched via airdropAirdropDistributing tokens for free to eligible wallets, usually to reward early users, bootstrap a community, or decentralise token ownership away from a small group of insiders at launch.Like a supermarket handing out free samples to people who already shop there. The samples cost the supermarket nothing to print. The goal is to convert casual shoppers into loyal customers by giving them something tangible to talk about.Read more → in January 2025. No presale. No ICOICOInitial Coin Offering. A token sale where a project sells tokens directly to the public, usually before any product exists. ICOs dominated 2017-2018 funding and are now mostly replaced by airdrops, IDOs, or fair launches.Like a company selling shares to the public before going public, except with no SEC oversight, no audited financials, and often no product at all. The 2017 ICO boom showed why those guardrails exist in traditional finance.Read more →. No private roundVCVenture Capital. Private investors who fund projects at an early stage in exchange for equity or token allocations. VC rounds are typically pre-launch, at steep discounts to any future public price, with multi-year vesting.Like angel investors in a startup who buy shares before the company goes public. They take more risk because the company might fail, so they get a better price. Once the company IPOs they can sell, and the public market pays whatever price it thinks is fair.Read more →. No VC allocation. Voorhees self-funded the entire operation. Genesis supply was 100 million VVV, distributed as follows:

• Airdrop to Venice users: 25% (100,000+ eligible wallets, 45-day claim window)
• Airdrop to crypto AIDeAIDecentralised AI. An umbrella term for blockchain-based projects that build AI infrastructure (compute, data, inference, models, agents) without a single central provider controlling the system.Like the difference between streaming a movie from Netflix and sharing it via BitTorrent. Netflix is fast and polished but one company controls what you can watch and what you pay. BitTorrent is messier but no single operator can shut you out.Read more → communities: 25% (Virtuals, Aero, Degen, aixbt, Morpheus and others)
• Company operations: 25% (no published vestingVestingA schedule that locks up tokens allocated to insiders, investors, and team members, releasing them gradually over months or years. Vesting prevents insiders from dumping on public buyers immediately after launch.Like a new employee's stock options at a startup. You don't get all the shares on day one. They unlock over four years so you stick around and do the work rather than cashing out and leaving.Read more → schedule)
• Team: 10% (25% unlocked at launch, remainder vesting over 24 months)
• Incentive fund: 10% (developer grants of $5,000-$100,000)
• LiquidityLiquidityHow easily a token can be bought or sold without moving the price. High liquidity means you can enter or exit large positions quickly at the quoted price. Low liquidity means even small trades can swing the market.Like the difference between selling a house and selling a share of Apple stock. The house might be worth more on paper, but finding a buyer at that price takes weeks. The Apple share converts to cash in one click.Read more →: 5% (deployed to Aerodrome DEXDEXDecentralised Exchange. A trading venue where token swaps happen entirely through smart contracts, with no central operator holding user funds. The largest DEXes are Uniswap, Aerodrome, Raydium, PancakeSwap, and Curve.Like a self-service vending machine that lets you swap one type of coin for another. The machine sets the exchange rate based on its current stock, anyone can deposit coins to refill it, and there's no clerk behind the counter.Read more → on Base)

The airdrop distribution was genuinely generous. When 32.6 million tokens went unclaimed after the 45-day window, Venice burned them rather than reclaiming them. That burnBurnPermanently removing tokens from circulation by sending them to an address that no one controls. Burns reduce total supply, which (all else equal) makes each remaining token worth more of the network's value.Like a company buying back its own shares and shredding them. The company's total value stays the same, but each remaining share now represents a slightly bigger slice of that value.Read more →, valued at roughly $100 million at the time, removed about a third of genesis supply permanently. Combined with ongoing revenue buybacks, the burn address now holds 33.72M VVV as of April 2026, which Venice frames as 42.45% of effective supply (total minted minus burns). BaseScan confirms the 0x0 balance on the VVV contract (0xacFE6019Ed1A7Dc6f7B508C02d1b04ec88cC21bf).

EmissionsEmissionsNew tokens created and distributed by a blockchain protocol over time as rewards to validators, stakers, or miners. Emissions fund network security and participation at the cost of diluting existing holders.Like a company that pays employees partly in newly printed shares. Every year the total number of shares goes up, which means existing shareholders own a slightly smaller slice of the same company unless the company grows faster than the printing.Read more → have been cut aggressively through five reductions in 13 months: 14 million → 10 million → 8 million (October 2025) → 6 million VVV annually (February 2026). That is a 57% reduction from the original rate. With the DIEM restructure, stakers now receive 100% of emissions (80% when sVVV is locked for DIEM). Monthly revenue buybackBuybackUsing protocol revenue to purchase tokens on the open market, usually to burn them or return them to a treasury. Buybacks convert business income into upward pressure on the token by reducing circulating supply.Like a public company using profits to repurchase and retire its own shares. The cash leaves the company's balance sheet, the share count drops, and every remaining shareholder owns a slightly bigger slice of the same business.Read more → burns have been active since December 2025 and are scaling with the business. The venice.ai/token dashboard shows the monthly burn climbing from roughly $50K in December 2025 to about $123K in April 2026, the largest monthly buyback to date. That trajectory matters more than the absolute number: it tracks subscription revenue growth in near-real time.

Measured against new emissions, the buyback isn’t net deflationary yet. April 2026 burns covered 3.3% of monthly emissions, down from 8.4% in December 2025. USD spend is rising while VVV-count coverage falls, because the token price has outpaced the buyback budget. The burn-vs-emissions strip on the chart at the top of this review tracks the trajectory month by month. Two more emission cuts are scheduled (5M annual from 1 May 2026, 3M from 1 July 2026), which will compress the gap before any further increase in buyback dollars.

The launch was not clean. On day one, two Aerodrome contributors purchased VVV before public announcements, turning $50,000 into over $1 million within an hour. Aerodrome flagged the trades within 30 minutes and suspended both individuals within three hours. VVV dropped 50% on the allegations. The insider trading was external to Venice, but the damage to launch credibility was real.

VVV is down roughly 75% from the all-time high of $22.58, reached one day after launch. Listed on Coinbase, Kraken, KuCoin, Bybit, Gate.io, OKX and Binance futures. See live data above for current pricing.

How to participate

Use the platform. The simplest entry point. Free tier requires no account. As of April 2026, Venice runs four tiers. Pro at $18/month (10% off annual) for unlimited text in the web app, 1,000 daily image generations, advanced model access (including Claude and GPT-5.x), voice mode and Memoria, plus 100 monthly credits for video/music/premium models and pay-per-use API. Pro Plus at $68/month adds 7,500 monthly credits with two-month credit banking (unused credits carry forward). Max at $200/month gives 22,500 monthly credits, three-month credit banking, and the highest image generation limits. 100 credits equal $1 USD. Pro does not include direct API access; API usage is billed separately via purchased credits, DIEM staking, or x402 USDC pay-per-request (see below).

Stake VVV. Staking VVV earns emissions-based yield (nominal APY ranges from 20-72% depending on the period), but this yield comes from newly minted VVV, not protocol revenue. Real yield depends on whether VVV holds value against the inflationInflationThe annual rate at which new tokens are created and added to the circulating supply. Most networks use inflation to pay validators, stakers, and infrastructure providers from freshly minted tokens rather than real revenue.Like a landlord who raises the rent every year. If your salary goes up at the same rate, you break even. If it doesn't, you get poorer without noticing, because the number on your payslip hasn't changed but the ground under it has shifted.Read more →. Staking alone doesn’t give you inference access. Seven-day unstaking cooldown.

Lock and mint DIEM. To access inference via tokens, lock your staked VVV (sVVV) into the DIEM escrow on the Venice token dashboard. Each DIEM represents $1/day of perpetual, renewing API credit. Stake DIEM (minimum 0.1) to activate access to all models on Venice including Claude, GPT-5 and Gemini. The mint rate rises exponentially as DIEM supply approaches the target set by Venice; supply is now past target, so the curve is steep. Early minters paid a fraction of today’s rate. While your sVVV is in escrow, you still earn 80% of standard staking yields. To exit, burn the same amount of DIEM you minted to unlock your original sVVV (partial unlocks permitted). DIEM also trades on Aerodrome (Base) and was added to Coinbase Exchange in late April 2026 (DIEM/USD pair). You can buy it directly instead of minting, though the secondary market price has risen sharply through April 2026. Over 8.26 million VVV is currently in escrow for DIEM, about 10% of effective supply. A recent capacity optimisation means DIEM allocation is now divided among active API users (those with an API call in the last seven days) rather than all stakers, resulting in a 14x capacity increase per token for active users. For the full mechanism, mint curve maths and short-squeeze risk vector, see our DIEM tokenomics deep-dive.

Getting started. The fastest path to Venice is the web app at venice.ai. No account required for the free tier (10 text prompts, 15 images per day). Pro at $18/month gets you unlimited text with advanced models including Claude, GPT-5 and Gemini, 1,000 daily image generations, voice mode and Memoria. Heavy video/music users will want Pro Plus ($68/mo) or Max ($200/mo) for the larger monthly credit allocations. For developers, the API is OpenAI-compatible (change one base URL) with four payment paths: prepaid credits (pay per token), DIEM staking, direct Pro tier allocations, or x402 USDC rails. Available API models include DeepSeek V3, Llama 4 Maverick, Qwen 3.5, Kimi K2.5 and others across text, image, code, and audio.

Build on the API. The OpenAI-compatible API is the developer on-ramp. The $27 million incentive fund has run two cohorts (110+ applications in Cohort 1, roughly 30 semifinalists in Cohort 2). Grants range from $5,000 to $100,000 in VVV. In April 2026, Venice added x402 support on the chat/completions endpoint: agents and developers can pay per request in USDC on Base, with a $5 minimum top-up. If you hold staked DIEM, your DIEM allocation is consumed before any USDC is debited. x402 is a credible agent-native payment rail but it does not create direct VVV demand; its contribution to the token economy is indirect, via the subscription and API revenue that funds the monthly buyback burn.

Honest assessment

What works

Venice is the best consumer-facing private AI inference product available today. That is not a high bar, but Venice clears it convincingly. Privacy architecture is substantively well-considered. Staking creates real utility demand rather than empty governance theatre. No VC allocation means no unlock schedule looming over the token. Product velocity has been impressive: text, image, video, audio, music, code, web search, voice mode and Memoria all shipped within 18 months of launch. And the aggressive supply tightening (33.72M VVV at the burn address, 42.45% of effective supply per venice.ai/token, emissions cut 57% from 14M to 6M annually, monthly buybacks growing every month since December 2025) shows a team that actually understands tokenomics.

What doesn’t work yet

Governance. Venice describes token-weighted governance but hasn’t deployed any voting interface. No Snapshot. No Tally. No on-chain governanceDAODecentralised Autonomous Organisation. A way to coordinate decisions and manage a treasury using token-weighted voting instead of a traditional company structure. Token holders propose and vote on changes directly.Like a shareholder-run company where every shareholder can vote on every decision, the votes are public, and the company can't do anything the shareholders don't approve. The coordination is messier than a normal company but nobody has unilateral control.Read more → contracts. Every significant decision, including five emission reductions and the DIEM launch, has been made unilaterally by the team. There is a Featurebase upvote system for feature requests, but that is not governance. Venice is transparent about being a company rather than a DAO, which is honest, but it means VVV holders have no actual decision-making power.

The risk

Voorhees is the single point of failure. Sole funder, CEO, primary public face of a roughly six-person company. No succession plan. No external investors. No board. Two SEC settlements on his record (SatoshiDice in 2014, ShapeShift in 2024). The ShapeShift precedent is instructive: Voorhees built a centralised crypto exchange, hit regulatory walls, then pivoted to a DAO structure. Venice’s trajectory could follow a similar path.

The cybersecurity angle is the less obvious risk. Security publications (Certo, CyberPress, CPO Magazine, Infosecurity Magazine) have flagged Venice as a tool for generating malware. One real-world crime credibly traced to Venice-generated content could trigger a regulatory crisis. The UK free tier is already blocked under FSMA. Other jurisdictions will follow.

The closed-source proxy is both the product and the vulnerability. If it goes down, the entire service stops. If it is compromised, prompts in Private and Anonymous modes are exposed (E2EE prompts remain encrypted even if the proxy is compromised, which is the point). No independent privacy audit of the proxy itself has been completed despite being the third-most-requested feature from users (334 votes on Featurebase). The March 2026 TEE/E2EE launch partially addresses this: for users on those modes, the privacy guarantee is now hardware-attested and independently verifiable. For Private mode users (the majority), it remains trust-based.

My position

I hold VVV and DIEM. I stake VVV for API access and use Venice daily for both the web interface and API calls. I find the product genuinely useful. I also recognise that I am trusting a roughly six-person company with my inference privacy, and that trust isn’t the same as a cryptographic guarantee.

Freedom Score: 57/100

C grade. Full methodology at Freedom Score Methodology.

Infrastructure decentralisation (9/20): GPU compute sourced from Akash, Hyperbolic, Prime Intellect, NEAR AI Cloud, and Phala Network. Multiple independent providers give some hardware distribution. But Venice still runs a centralised, closed-source proxy that routes all requests. Single point of control and failure. Users can’t choose their GPU provider. Roughly six employees operate everything.

Governance decentralisation (5/20): Token governance described but not deployed. No voting interface, no proposal system, no quorum mechanism. Voorhees is sole funder, CEO and primary decision-maker. Five emission reductions and the DIEM launch were unilateral team decisions. Venice is transparent about being a company, not a DAO, which is at least honest.

Token distribution fairness (11/15): No presale, no ICO, no VC allocation. Self-funded. 50% of genesis airdropped. 32.6 million unclaimed tokens burned rather than retained. Team allocation 10% with 24-month vesting. The insider trading incident was external (Aerodrome contributors, not Venice team). Company operations allocation (25%) has no published vesting schedule, which is a gap.

Censorship resistance (11/15): The strongest censorship resistance of any consumer AI product today. Pro subscribers can disable all content filtering. Open-source models that can’t be centrally recalled. But Venice controls the proxy and can theoretically blockBlockA batch of transactions added to a blockchain at a set interval. Each block cryptographically links to the previous one, creating an append-only chain that can't be rewritten without redoing all the work since.Like a page in a ledger. Every page has a fixed number of entries, every page references the previous page, and once a page is filled and signed off it can't be edited without visibly invalidating every page that came after. The chain is just a very long series of these sealed pages.Read more → users, models or content types. Free tier already blocked in the UK. The resistance is a business decision, not an architectural guarantee.

Data sovereignty (12/15): No server-side storage of prompts or responses in any mode. Conversation history in local browser storage only. Proxy strips identifying information before forwarding to GPUs. Strong privacy architecture with minimal metadata collected and crypto payment option. In Private mode, GPU providers see individual prompts but not user identity or full conversation history. Privacy relies on architectural isolation and Venice’s policy rather than cryptographic guarantees. No independent privacy audit exists despite being the third-most-requested feature. Competitors (Phala, Secret Network) offer stronger cryptographic privacy at the computation layer.

Open source and transparency (9/15): Uses open-source AI models. Smart contracts verified on BaseScan. CLI and agent runtime are open source. But the core inference engine, proxy layer, and platform code are all closed source. Venice’s privacy claims cannot be independently verified because the proxy code is not public. No independent privacy audit despite it being the third-most-requested feature. Revenue and financials not disclosed.

Path to improvement

Three changes would materially increase Venice’s score:

1. Expand E2EE to all models and free tier. The March 2026 E2EE/TEE launch is a genuine breakthrough, but it’s limited to Pro subscribers and a subset of models. Extending hardware-attested privacy to free tier users and the full model catalogue would make Venice’s privacy guarantee universal rather than tiered.
2. Open-source the proxy layer. The centralised, closed-source proxy is the primary freedom score deduction. Open-sourcing it would not eliminate the centralisation (Venice would still operate the instance), but it would allow independent verification and, eventually, self-hosted alternatives. The TEE/E2EE modes reduce the proxy’s trust requirements, but the proxy still handles routing for all modes.
3. Deploy governance. Ship a token-weighted voting interface, even for non-binding signal votes. Venice describes governance as a feature of VVV but hasn’t built any of the tooling. Deploying Snapshot or an equivalent would convert a marketing claim into a functional mechanism.

Returns Score: 68/100

B grade. Full methodology at Returns Score Methodology.

Token utility (16/20): VVV has a well-designed utility loop. Staking VVV earns emissions-based yield, and holding sufficient staked VVV grants pro account access. To get inference from tokens, you lock staked VVV to mint DIEM, then stake DIEM. Your inference capacity is proportional to your share of staked DIEM. Each DIEM represents $1/day of perpetual API credit against any model. This creates genuine lock-up demand from heavy API users. The April 2026 x402 launch adds a fourth payment rail (USDC on Base) but routes through any staked DIEM allocation first, so DIEM utility is preserved. The utility is practical and directly tied to product usage rather than abstract governance rights that nobody exercises.

Value accrual (12/20): Venice’s monthly revenue buyback burns have been active since December 2025 and are scaling with revenue. The venice.ai/token dashboard shows the monthly burn climbing from roughly $50K in December 2025 to about $123K in April 2026, the largest single-month buyback to date. The April 2026 introduction of Pro Plus ($68/month) and Max ($200/month) subscription tiers materially expands the revenue surface that funds the buyback: heavier users now have a reason to pay 4x or 11x the Pro tier price. x402 USDC payments add a new developer revenue channel. The honest caveat: x402 is USDC-based and does not create direct VVV demand. Value accrual remains indirect, mediated by the buyback mechanism. Venice still doesn’t publish revenue, so the buyback size has to stand in as the proxy measurement.

Supply dynamics (15/20): Genesis supply was 100 million VVV with no presale or VC allocation, which is a strong starting point. The 32.6 million unclaimed airdrop tokens were burned rather than reclaimed, a credibility-building move worth approximately $100 million at the time. Emissions have been cut five times in 13 months: from 14 million down to 6 million VVV annually (October 2025: 10M → 8M; February 2026: 8M → 6M), a 57% total reduction. Monthly revenue buyback burns have been active since December 2025, creating a genuine two-sided tightening: less new supply entering, existing supply being permanently removed. BaseScan confirms 33.72M VVV at the burn address (42.45% of effective supply per venice.ai/token). The problem remains the company operations allocation: 25% of genesis supply with no published vesting schedule. Voorhees self-funded the project, which means no VC overhang, but the company can sell its allocation at any time without constraint. Annual inflation is now approximately 7.6% at the reduced 6M rate, moderate but not negligible.

Revenue sustainability (13/25): Venice reports 1.3 million registered users and 45 billion LLM tokens processed daily, with Warden Protocol alone accounting for roughly 1.5 million daily chats according to the company. The subscription model now covers four tiers (Free, Pro $18/mo, Pro Plus $68/mo, Max $200/mo) plus two API-specific rails (prepaid credits, x402 USDC on Base). That tier expansion is the most significant commercial move since launch. It signals that heavy users were willing to pay substantially more than Pro but had nothing to pay for, a common indicator of product-market fit. x402 provides an agent-native rail without requiring developers to hold VVV, lowering friction for autonomous agent integrations. Venice still hasn’t disclosed hard revenue figures, but the monthly buyback growth ($50K → $123K between December 2025 and April 2026) is a verifiable proxy. The sustainability of a roughly six-person operation running infrastructure of this scale remains an open question.

Liquidity and access (12/15): VVV is well-covered for a token of its market cap. Listed on Coinbase, Kraken, KuCoin, Bybit, Gate.io, OKX and Binance futures, with strong daily volume. The Base-native DEX liquidity through Aerodrome provides a solid decentralised trading option. You can enter and exit positions of reasonable size without difficulty. The Aerodrome insider trading incident on launch dayTGEToken Generation Event. The moment a project's token first becomes tradeable. TGE is when vesting clocks usually start, when liquidity hits exchanges, and when public price discovery begins.Like the IPO day for a startup. Everything that happened before TGE was private valuations and paper agreements. Everything after is the public market deciding what the thing is worth in real time.Read more → was a credibility hit but did not structurally impair liquidity. Exchange coverage is strong relative to comparable DeAI tokens.

Path to improvement

Three changes would materially increase Venice’s returns score:

1. Publish revenue figures. The buyback burn mechanism only matters if holders can verify the numbers. Quarterly revenue disclosure (even aggregate figures without client-level detail) would let the market price VVV on fundamentals rather than trust. Every other dimension of the token economy is transparent; revenue is the conspicuous gap.
2. Publish the company operations vesting schedule. The 25% company allocation with no vesting constraint is the largest unresolved supply risk. A time-locked, on-chain vesting schedule would remove ambiguity and align the company’s token management with holder interests.
3. Grow the DIEM adoption base. DIEM is the most elegant lock-up mechanism in the DeAI space: it converts VVV staking into perpetual API credit. But adoption depends on API usage growth. Expanding the developer community, running more grant cohorts, and building integrations that drive sustained API demand would increase DIEM minting, which locks VVV and tightens effective supply.

Score change log