Every figure, name and quotation the finished picture puts on screen, chased to a primary source.
Venice Uncensored is a Venice AI model built with the Dolphin AI team. The Venice model page credits both. The Hugging Face card for the model is published under the AskVenice organisation.
It is based on Mistral Small 24B Instruct and on Dolphin Mistral 24B Venice Edition.
The Hugging Face card gives the lineage explicitly: built on
mistralai/Mistral-Small-24B-Base-2501, fine tuned through
mistralai/Mistral-Small-24B-Instruct-2501, and further fine tuned from
dphn/Dolphin-Mistral-24B-Venice-Edition.
24 billion parameters.
Stated as 24B params on the Hugging Face card and as a
“24-billion-parameter open-weights text and vision model” on the Venice model page.
128K context window, on Venice Uncensored 1.2. The Venice model page lists a 128K token context window. Version 1.2 is dated 1 April 2026 and is published under Apache 2.0.
The card describes the model with exactly the three words the video quotes, and gives each a one line definition:
It also states that the model “has no hard-coded alignment, it relies on you to tell it how to behave.”
The Dolphin card behind it makes the same argument in its own words: “Dolphin does not impose its ethics or guidelines on you. You are the one who decides the guidelines,” and “the system prompt is what you use to set the tone and alignment of the responses.”
Free plan: 10 text prompts per day. Pro plan: $18 per month. The Venice pricing tiers list a Free plan at $0 per month including “10 text prompts per day”, and a Pro plan at “$18/mo” with unlimited text prompts and 1,000 images per day. Higher tiers exist above it (Pro Plus at $68 per month, Max at $200 per month) and are not discussed in the video.
API token pricing for Venice Uncensored 1.2 is $0.20 per million input tokens and $0.90 per million output tokens. The video does not state these, and they are recorded here only because the price of running it is the obvious next question.
It implements the OpenAI API specification. The documentation states: “Venice’s API implements the OpenAI API specification, ensuring compatibility with existing OpenAI clients and tools,” and describes integrating “using the familiar OpenAI interface”.
The surface the video lists. The published documentation index covers, as separate endpoint groups: image generation and editing; text to speech and speech to text; music and sound effects; text to video and image to video; embeddings; character personas; document processing (“Extract text from PDF, DOCX, XLSX, and plain text files”); and web tools covering search, scraping and text parsing. Chat completions carry text, and code is produced through them rather than through an endpoint of its own.
Venice specific request behaviour, including enable_web_search and
enable_web_scraping, is set through venice_parameters on an otherwise standard
request.
Venice documents four tiers, each building on the one before it:
E2EE — “Your client encrypts the prompt before sending it. Venice relays ciphertext, and only the verified TEE decrypts it.”