Venice AI's Uncensored Updates

This week we’re checking back in on another privacy-focused AI — Venice from Erik Voorhees. 

The platform, built for both users and developers, launched in 2024. Since we last covered the project in January, the team has added a series of fresh features, including a mobile app, new AI models, and a new compute token, DIEM, which premiered last week. 

Whether you're looking for occasional private queries or planning to build uncensored applications, it’s worth learning about Venice’s latest offerings. Let’s dive in.

A Refresher on Venice’s Core Features

As with all the more “traditional” AIs on the market today, Venice provides text, image, and code generation, but without the guardrails you'd encounter elsewhere. 

The default model used for text is Venice Uncensored (Dolphin Mistral 24B Venice Edition), a model built alongside Dolphin, an AI development team, that reportedly excels at “creative writing, role-playing, [and] philosophical discussions.”

While the platform currently supports five different models, only two are available for “free” (more on this later). There is also document analysis for PDF and other files, letting you extract summaries or answer questions about documents without external processing.

Venice has also added a Social Feed for discovering and sharing prompts created by the community, as well as Characters, which are custom AI personas built and shared by users.

Underpinning all this is Venice’s privacy system, a decentralized architecture that routes encrypted queries through a network of GPU providers rather than centralized servers. As a result, conversations aren't permanently stored on Venice's servers but rather kept in your browser's local storage, encrypted to prevent direct observation.

While good for privacy, this comes as the cost of convenience since it means if you log into Venice from a different device, your chat history won't carry over since it was never synced to a central database. Venice claims to track only basic events like sign-ins or new chat starts for product feedback, never logging actual prompt content or model responses.

Venice & Crypto’s Escalating AI Ambitions on Bankless

Crypto’s latest AI coin, $VVV, arrives with a billion-dollar bang on Base.

BanklessDavid Christopher Jan 28, 2025 • 6 min read

While this approach certainly provides a higher level of security and privacy, know that you're still trusting that the GPU nodes processing your requests aren't logging data themselves. 

Tiered Access Levels

Venice structures access simply. The free tier requires no account or email, providing 25 text prompts and 15 image generations daily with access to two models: Venice Medium (Mistral Small 3.1 24B) and Venice Uncensored. These are models are great for testing the platform or handling occasional sensitive queries you'd rather not run through monitored services.

As for the Pro tier, it costs $18 monthly or $149 yearly, and it removes all generation limits and provides access to all five model categories. Beyond Venice Medium and Uncensored, Pro users can access:

• Venice Reasoning (Qwen QwQ 32B) for complex problem-solving.
• Venice Small (Mistral Small 3.1 24B) for balanced performance with vision capabilities.
• Venice Large (Qwen 235B) for technical analysis with a massive 256K token context window. 

Pro also includes unlimited prompts, uncensored image generation without watermarks, image editing, and Character creation — plus the ability to share prompts to the Social Feed. Pro users additionally get API access for app development and integration, plus some credits to play around with.

The New Venice Mobile App

Building on their web-based tools, Venice launched its mobile app for iOS and Android this month. 

While the platform was already available on mobile web, the app makes for a tighter, clean experience for accessing core features, all processed with local encryption and no central logging. Further, with the Social Feed for community prompts, it comes equipped with a level of discoverability that feels “at home” on mobile.

• For iOS users, you can download it from the App Store.
• For Android, you can download Venice from the Google Play Store.

DIEM: A Stable Approach to API Access

Equally as important as its app is Venice’s API, enabling programmatic access to its uncensored AI.

As a critical part of the platform’s offering, Venice launched a $27M incentive fund for building with its API, as well as last week’s launch of the secondary token, DIEM, to make API access more efficient and reliable.

DIEM, mintable only by locking staked VVV (sVVV), delivers perpetual $1 daily API credits when staked (minimum 0.1 DIEM), improving on Venice’s pre-launch model where API access came from VVV staking directly and was pro-rata — your credits fluctuated based on your share of the total staked pool, diluting with more participants and creating unpredictable capacity for devs. 

Now, VVV purely acts as “equity” in Venice, only earning staking rewards, while DIEM, mintable through VVV, provides the actual discounts for API usage, with a more reliable budgeted model. But, beyond just reducing costs for developers, this new dynamic unlocks interesting DeFi strategies for VVV stakers who may not be using Venice’s API. In this two-token system, you could:

• Mint DIEM and sell it for liquidity or LP it. While you’ll need DIEM to unstake your VVV, you will still earn staking rewards in VVV while your sVVV is locked up, though at a discounted rate.
• Leverage up on VVV by minting DIEM, selling it for more VVV, restake and repeat for amplified yields though know, once again, you’ll need to re-obtain DIEM to unlock your VVV.

If you’re interested in learning more about how to build with Venice’s API, you can access their docs here.

In the past few months, Venice AI has made significant strides in expanding the accessibility and functionality of its suite.

Updated models provide competitive performance across text, image, and code generation. The mobile app brings a consumer-friendly gateway to true on-device privacy without accounts or tracking. DIEM’s launch has also stabilized API access, while expanding DeFi functionality for Venice’s ecosystem overall. 

To quickly compare the project with last week’s focus, Nillion's private chat AI NilGPT, Venice offers a more comprehensive toolkit — five models versus one, image generation, code capabilities, mobile apps, and a mature token ecosystem.

NilGPT, still in its early stages, focuses purely on text with a simpler interface, though crucially it allows conversation syncing across devices through encrypted cloud storage — something Venice's local-only approach doesn't support.

For users wanting a straightforward ChatGPT alternative they can access from anywhere, NilGPT may prove more convenient. But for those needing the full generative AI suite, Venice provides the deeper feature set.