Uncensored DeAI on Mobile
View in Browser
Sponsor: KGeN — KGeN is building the Verified Distribution Protocol for the next Web3 wave.
📸 Market Snapshot: Wayfinder’s $PROMPT stole the spotlight this week as Monday’s Open Alpha launch triggered a massive surge. The token climbed over 200% in initial hours and maintained net gains of ~70% as the week progressed, despite crypto’s overall rocky week.
To quickly refresh on Wayfinder, it’s a protocol enabling autonomous agents, whether they be for yield farming, gaming, or otherwise, to navigate and operate across multiple chains. The protocol’s built by Parallel Studios, the creators of Echelon Prime, as an extension of the project’s ecosystem. $PROMPT serves as the native utility and governance token, powering agent operations, rewarding contributions, and enabling holders to influence protocol decisions through staking and voting mechanisms.
The Open Alpha brings core AI agent functionalities like:
- Transaction Agents for handling transfers
- Contract Agents for smart contract interactions
- Perpetuals Agents for derivatives trading
- Prediction Agents for market forecasting
Further, as is all the rage these days, the platform chose to integrate Hyperliquid for trading. This taps into not only the chain’s liquidity, but its popularity—which could pay dividends as everyone continues to hunt for an agent that can earn them money trading while they do nothing. If you find one, let me know.
The Open Alpha also features tools like a Discovery page highlighting Top Gainers, Pump.fun graduates, yield opportunities, and recent crypto news — market intelligence, so to speak. Users get real-time market views, simulation previews, and transparent transaction summaries.
It does come in the form of a tiered system: The Free Plan lets you try all major features with message & tx caps. The PRO Plan offers higher usage limits and enhanced access for power users.
Meanwhile, Bittensor subnet tokens have continued showing pockets of outperformance: Bitcast (SN 93) surged 91% as a decentralized creator economy rewarding brand-related content, Ridges (SN 62) moved 68%, focusing on building engineering agents, and Dippy (SN 11) added 46% powering roleplay AI companionship.
Additional smaller caps like Zeus (SN 18) and Mantis (SN 123) put up ~55% gains alike. While expanding to Base has likely provided these tokens some extra juice and demand, the small size of Bittensor’s “economy” likely plays a larger role in keeping these afloat despite choppy waters for the majority of tokens overall.
(All numbers above were recorded at the time of writing, Aug. 28th!)
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.
David 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.
Zooming Out
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.
Plus, other news this week...
🤖 AI Crypto
- 🔥 Bankr — activated gas-free trading on Base, Polygon, and Unichain
- Solana — introduced Solana Bench, a new testing system for gauging how well LLMs operate on the chain
- Vader — announced a pivot away from developing around Virtuals Genesis products
- Virtuals — rolled out support for liquid staking via $VVET
- Wayfinder — launched the Open Alpha of its AI agents platform
📣 General News
- 🔥 Google — unveiled Gemini 2.5 Flash Image, a.k.a. nano-banana, its most advanced image gen and editing model yet
- Meta — partnered with Midjourney to license the company's "aesthetic technology" in future products
- xAI — filed a lawsuit against Apple and OpenAI, alleging anti-competitive collusion over their ChatGPT partnership
📚 Reads
- 🔥 TJ Kawamura — Physical AI: The Next Frontier for Data
- Bethany Crystal — Why 95% of AI Pilots Fail
- Chakra Labs — Computer Use Agents Pt. II
KGeN is building the world’s largest Verified Distribution Protocol, a.k.a VeriFi — delivering AI & Human Data-driven intelligence through KAI. With over 200+ enterprise partners, and global human-verified data at scale, KGeN has achieved a massive $33.6M+ ARR as of July 2025.
In the latest Limitless episode, Josh and Ejaaz unpack why Google just seized the AI crown, starting with its new Gemini 2.5 Flash Image model delivering instant, ultra-cheap, character-consistent generations.
They also break down how Google’s full stack (Gemini, Veo, Genie, AI Studio, Notebook LM, etc.) feeds into a single accelerating model loop. Along the way, they hit Meta’s Midjourney aesthetics deal, the rise of AI companions across Instagram, and rumors of “Hypernova” glasses with wristband controls.
Finally, the duo preview Gemini 3, Apple’s next AI moves, and why September’s hardware season could redraw the AI leaderboard in real time.
Tune in to catch up on all this week's biggest intrigues! 👇