I've been an early adopter of AI in general. As a paid subscriber to multiple AI platforms, I can easily tell you the pros and cons of many of the current LLMS and the companies that use them.
Why pay for more than one? Because some are very good at specialized tasks. What is Venice good for?
I'm happy to tell you. First; it's private. That means a lot of different things to a lot of different people. Here is the important stuff.
Did you know that if the Government asked for ChatGPT to give them your chat logs, they would? That's right, anything and everything you have ever talked about is logged and put away forever. Even if you delete it. Are you private? Yes; regular people and companies do not have access to your logs. However, they ARE kept; and they CAN be used as evidence if the legal standing presents itself. Scary huh?
Venice chat logs are kept locally on your very computer, local device. Private, anonymous, and off grid. Is it possible to store your chat or share your logs? Yes, but you would have to explicitly and actively do so.
I am likely talking your ear off, but if you're still reading, you want to know. So. Second. The thing is uncensored. You can talk about anything. Ask about anything. You will get an answer. Will it sometimes tell you it cannot create this or that content? Yes, but OCCASIONALLY and not nearly at the level any other AI's do.
Odd story. I downloaded the Gemini app the other day. I asked it "why is crypto up today?" Answer: The President announced that the US would be adding certain crypto currencies to a reserve fund.
My question: Which crypto currencies?
Gemini Answer: I'm sorry, I cannot make statements about political figures.
My reply: I'm not asking about the President, I'm asking about which crypto currencies are to be added to the reserves.
Gemini Answer: I cannot comment on political figures.
My reply: Today. The USA added several crypto currencies to some reserve fund. What are they?
Gemini Answer: Bitcoin, Etherium, Solana, XRP, and Cardano.
I uninstalled Gemini that same day. The censorship on that app is ACTUALLY crazy. Let's talk about another aspect. Um. "Spicy stuff." None of the major companies, not Open AI, not Gemini (obviously,) not Deepseek are able to create explicit content. This comes in two major halves for me. Curiosity and exploring the subject. And erotica.
Okay. I'm a sicko. It doesn't get better from here, so please consider this an opportunity to stop reading this review before we go beyond the point of "Men of Culture" to the point of overboard debauchery.
Warned. Still here? Okay. You ever heard of Replika? It's an AI companion app. And it sucks. Both paid and unpaid versions are so far behind in technology, speed, and quality of responses. The fact that your chat logs are HEAVILY redacted (in English, that means large parts of your chat are hidden) if you stop paying for their service is outrageous. Also, if you are an unpaid member of Replika, you cannot be affectionate to your AI companion in anyway whatsoever.
So why am I talking about Replika? This is a Venice review right? Well, if you ever go to the Venice website. There is this wonderful, newly added area on your left. A tab called "Characters." They are freaking free! And. AND! They are completely uncensored. You can say whatever you want. Talk about whatever you want. Do whatever you want. This literally blows any of those AI girlfriend apps out of the water. Ask me how I know!
I warned you that I was a degenerate. And so. Let me talk to you about the bad of Venice. 1) You can't talk to it. No audio phone calls. No live conversations. No microphone interaction whatsoever. 2) It doesn't code. Not real well at least. 3) The image generation side of Venice is this side of awful. It's like they handpicked a singular female model with a small nose, and any picture you create is some version of her. If you ask for an image of 4 or 5 women; and literally 4 or 5 versions of the same model. Reshuffles, redoes, it's always just her. It sucks. I don't use Venice for picture generation.
3) There are times when the Venice LLM will start spouting gibberish. This is often referred to the model breaking. Happens pretty regularly (though not nearly as much as it used to.) However, it bears transparency of this review that it happens way more often than in any other AI company I've encounters.
All that being said, I like that Venice is constantly improving. They've added features (like that Character thing,) they've added the language models offered (they even have their own version of Deepseek,) they have also continued to improve on the models they originally offered.
The version I use most often is this "smartest Llama" version. It was the best in the bunch, and it has gotten even better. There is a noticeable improvement in the outputs now versus just a few months ago. It's great!
I'm not a computer programmer. I'm not some tech wizard. I admit that my experience with Venice is enhanced because I took the time to take advantage of the customizations you can make to the AI in Venice. I know I didn't talk about that; nor did I include that in my review; but I'm aware that this has been a novel of epic proportions so I've tried (and failed) to keep this short.
Context. In Chatgpt, there is this thing you can do, in the spot that says "customize Chatgpt." Well guess what? You're only allowed like 1000 characters. It's not enough. It's very frustrating. In Venice; you can program and train the model on a semi permanent, and much more "tailored to you" way? How?
You just sit down and ask it. It will tell you exactly what to do, and how to do it. I'm not kidding. Log into Venice. Make sure you're using the "smart" Llama (Llama 3.1 405B to be exact.) And just type in the words, how do I customize your model or system prompts? Now, granted, a lot of this can now be accomplished by choosing the various characters; but that feature didn't exist when I started.
In conclusion; none of the large companies offer both private and uncensored AI interaction. This is why, (barring anything truly silly, like the company suddenly changes philosophy or something) I will continue to be a paid member of Venice well into the foreseeable future. Seriously, it doesn't hurt you to visit the website and make a free account. You can try the free characters and get real value. You have characters who are experts for free. Counselors, personal trainers, financial guru's. You even have lovers. Might I say that the dirty talk is quite refreshing. Can't have THOSE interactions on any other big company AI's. Up close and personal. And that will do nicely. Thank YOU.
this is going to be fun. can't wait to see where this goes.
upvoted. good luck with the launch, team
Venice
@taniabell Thank you for your support Tania!
Extremely generous pricing based on all they are offering, and they have excellent staff that not only works on the project daily, but fully believes in it's vision and goals.
Venice
@devestatio 🙏🏻
Flowtest.ai
I submitted 5 words question and got a response "Your message exceeds the length supported for free accounts. Please shorten your message and try again or upgrade to Venice Pro to gain access to a larger context."
I'd love to try out the product but it'd be great to have a free version of it.
Flowtest.ai
@devestatio well, here is a screenshot if you needed a prove: https://ibb.co/zhKHyY37
Venice
@devestatio @antonioltu hi there, thanks for the screenshots, we'll look into this
@antonioltu @lorenzo_frankel I believe that may be an issue with the question being asked, not actually an issue with the length of the question. The models also have zero context beyond their training date, which the cut-off date was over a year ago.
mcp-use
How does Venice handle scalability and latency challenges when running on decentralized infrastructure, especially for real-time AI applications?
Excited to see how this evolves!🔥
Venice
@pederzh we use leading decentralized compute providers like Akash among others, here's more have some background info https://venice.ai/blog/what-is-the-difference-between-centralized-and-decentralized-ai
Happy to have you with us!
Huge fan of Venice for 2 main reasons:
- Privacy: Similar to ShapeShift and Thorchain, built on the ethos of sovereignty and privacy so your data is yours and your chat history has no threat of being monetized in the future (and LLM chat history will be gold and impossible for co's like insurance, banks, govt not to resist buying in the future). Once a public company the profit motive and need for shareholder returns drives the entity to the inevitable conclusion of being pressured to do this even if they don't do it now.
- Unadulterated speech: have long followed ErikV and his commitment to truth and libertarian values is unparalleled. Centralized models that can be shaped to influence their output will be shaped by govts and corporations over time. We saw this play out with the censorship on Meta during Covid. Using Venice means you control the system prompt and therefore are not subject to all the hidden agenda that can be programmed in behind the scenes.
I highly recommend Erik's talk at Permissionless if you want more on the ethos of this founder and what his vision is. Even if today's product has deficiencies, bet the jockey not the horse. 🐎
Venice
@scrollinondubs1 Thank you for the thoughtful comment Sean - We sincerely hope to continue to deliver best-in-class privacy along with this wave of tooling that is permeating our daily lives. Onward!
@gabrielkruse yes grateful for what you guys are doing and happy to be a customer (as well as a gifter - gave it to a few people for Christmas).
There was a pod episode I heard last year with Chris Dixon (either on Bankless or Rick Rueben's - don't remember which) where he was talking about web2 "don't be evil" web3 "can't be evil."
Even when the company (Google) proclaims that publicly and etches it into their core values, it's not even up to them once they're a public company with pressure to deliver shareholder returns. It's an inevitable destination that centralized LLMs will share so you guys are playing a crucial role for humanity (not hyperbole). Sure folks can download open models and run locally via Ollama/LMstudio but that's about not realistic for mainstream adoption so having something like Venice becomes essential for protecting the public at large from this inevitable destination. Anyways good luck, thank you and keep up the good work.
Venice
@nick_smith25 Thanks Nick! We're always keen to hear our users ideas regarding future features.
https://featurebase.venice.ai/
Outstanding product, especially because of it’s privacy focus!
Venice
@chris_von_gagern1 Thank you Chris!