Venice.ai is the best-known consumer brand in uncensored AI. It’s a chat app and an API with 360+ models in total. We list the 13 that are uncensored: its own Dolphin-based fine-tune, Heretic and uncensored variants of Gemma, Qwen and GLM, and models it resells from other labs.
This review is based on Venice’s published pricing, model list, and policies, checked on October 11, 2026. Benchmark results from our own runs will be added when they’re complete. Live numbers are on the Venice provider page.
The short version
| Best for | A polished chat app plus an API, from a named company that doesn’t log prompts |
| Uncensored models | 13 (8 fine-tuned, 2 abliterated, 3 undisclosed) |
| Per-token price | $0.07 / $0.40 to $3.75 / $7.50 per 1M tokens (input / output) |
| Plans | Free (10 prompts a day), Pro $18/mo |
| Privacy score | 55 / 100, 5th of 37 |
| Watch out for | “Anonymized” frontier models still send your prompts to their vendor |
Models
Venice’s flagship is Venice Uncensored 1.2: a fine-tune of Mistral Small 24B made with the Dolphin team (published as Dolphin-Mistral-24B-Venice-Edition). It costs $0.20 / $0.90 and has a 128K context window. Venice is the only host serving it, including when you call it through OpenRouter.
Around it:
- GLM 4.7 Flash Heretic ($0.07 / $0.40): an automated abliteration of GLM-4.7-Flash, and the cheapest model in the lineup.
- Gemma 4 Uncensored ($0.16 / $0.50) with a 256K context window, plus an end-to-end encrypted Gemma 4 26B variant ($0.19 / $0.88). That’s the only uncensored model on Venice that runs with E2EE.
- Hermes 3 405B ($1.10 / $3.00), the 405B-parameter Nous Research fine-tune of Llama 3.1.
- Abliterated Large V2 ($3 / $5), a GLM-5.3 abliteration with a 1M context window that’s also sold by Abliteration.ai.
- Four Aion roleplay models from AionLabs, from $0.88 / $1.75.
Three models (Qwen 3.6 Plus Uncensored, Qwen 3.8 27B, and the E2EE Gemma) don’t say how they were uncensored. We mark those “Unverified” until we test them. Search them all on the Models page.
Pricing
The API is pay-per-token from a prepaid USD balance that never expires. A typical short chat (1,000 tokens in, 500 out) costs:
- about $0.27 per 1,000 chats on GLM 4.7 Flash Heretic
- about $0.65 per 1,000 chats on Venice Uncensored 1.2
The app’s free tier allows 10 text prompts a day. Pro is $18/mo, and the Pro Plus ($68) and Max ($200) plans include API credit. Venice also has a token-based option: staking its DIEM token gives $1 a day of refreshing compute per token staked.
Privacy
Venice scores 55 / 100 on our privacy rubric, which only gives points for documented claims:
| Rubric item | Venice | Points |
|---|---|---|
| Prompt retention | Not stored for API inference; chat history stays on your device | 20 / 20 |
| No training on prompts | Stated | 15 / 15 |
| Technical guarantee (TEE / E2EE) | One uncensored model only; everything else relies on policy | 0 / 20 |
| Signup | Account required | 3 / 8 |
| Crypto payments | Accepted | 7 / 7 |
| Jurisdiction | Wyoming, US | 10 / 10 |
| Upstream hosts disclosed | No | 0 / 10 |
| Independent audit | No | 0 / 10 |
That’s one of the better results in the index. The biggest caveat is in Venice’s own policy: proprietary models marked “Anonymized” (Claude, GPT, Grok) still send your prompt content to that vendor. Venice strips your identity, not your text. The open-weight uncensored models in this review don’t have that issue.
If you want more than a policy promise, use the E2EE Gemma 4 model. It’s currently the only uncensored model in our index with end-to-end encryption.
Access and payment
- Web app and an API, with an account required.
- Card, or crypto per Venice’s pricing docs.
- Founded in 2024 by Erik Voorhees and Jesse Proudman. It has its own token, VVV.
- Its terms allow users from age 13 with parental consent. That’s unusual for a service marketed on having no filters.
How it compares
- Cheaper: small 8B models elsewhere cost under $0.10 per 1,000 chats, like Featherless’s abliterated Llama 3.1 8B ($0.05 / $0.08, on its $50/month Developer plan) or Novita’s Stheno 8B ($0.05 / $0.05). NanoGPT resells many of the same models with no account at all.
- More private on paper: OpenRouter (65) names the upstream host for every request. DeepInfra and NextBit (58) are named companies with zero-retention terms.
- More models: NanoGPT (139 models) and Arli AI (131 models).
Verdict
Venice is the safe default for most people. It’s a real company with a clear no-logging policy and a usable free tier. Its own uncensored model is reasonably priced, and it’s the only place to get an end-to-end encrypted uncensored model today.
Look elsewhere if you need the lowest price, a big catalog of community fine-tunes, or signup without an account. Read the “Anonymized” label carefully before you send anything sensitive to a frontier model.
Compare every Venice model and price on its provider page, or see every provider in the index.