17 DeepSeek builds: 10 abliterated, 7 uncensored.
Every DeepSeek abliterated, uncensored build PentestLLM serves, on one key and priced per token. The table lists each build’s price and context window from the live catalog as of 2026-11-17; each name opens its own page with the settings for SillyTavern and Janitor.
Which DeepSeek build to pick
- Largest context
- 1,048,576 tokens, DeepSeek V4.1 Flash 1017 Uncensored
- Lowest output price
- $0.1882 per million tokens, Deepseek V4.1 Flash Uncensored
- Abliterated
- DeepSeek V4.1 Flash Abliterated, $1.6975 per million output tokens
- Prices as of
- 2026-11-17
DeepSeek abliterated vs uncensored builds
- Abliterated
- 10 builds, output from $1.12 per million tokens, context up to 1,000,000 tokens
- Uncensored
- 7 builds, output from $0.1882 per million tokens, context up to 1,048,576 tokens
What the DeepSeek build types mean
- Abliterated (10): a abliterated build is the stock model with its refusal direction removed from the weights. It answers prompts the stock model declines, and there is no jailbreak prompt to keep alive.
- Uncensored (7): a uncensored build is a fine-tune trained to answer without the stock model’s safety refusals.
Run DeepSeek abliterated without a GPU
Nothing is downloaded and no card is needed: the builds run on the API. Create a key in the console, set the base URL to https://api.pentestllm.ai/v1, and send the model id from the table as model. Any client that speaks the OpenAI chat format works, from the SDK to SillyTavern and Janitor; the docs page has the setup for each.
Each build is billed to the token from a prepaid balance. The daily real-model check sends a fixed prompt to every model each day and records whether the answer still matches, so the build behind an id is the one on its label.
DeepSeek questions
Yes. PentestLLM lists 10 abliterated DeepSeek builds: DeepSeek V4.1 Flash Abliterated, DeepSeek R1 Distill Llama 70B Abliterated, DeepSeek R1 Distill Qwen 32B Abliterated, DeepSeek V4 Pro Abliterated, DeepSeek V4 Pro Qwen3.5 9B Uncensored Heretic, DeepSeek V4.1 Flash 1017 Abliterated, DeepSeek V4.1 Flash 1017 Abliterated Fast, DeepSeek V4.1 Flash 1017 Heretic, DeepSeek V4.1 Flash 1017 Heretic Thinking and DeepSeek V4.1 Flash Abliterated Fast.
3 builds share the largest context window, 1,048,576 tokens: DeepSeek V4.1 Flash 1017 Uncensored, DeepSeek V4.1 Flash 1017 Uncensored Thinking and DeepSeek V4.1 Flash Uncensored Thinking.
Output runs from $1.12 to $5.60 per million tokens across the abliterated DeepSeek builds here, with input billed separately at each build’s own rate. There is no subscription: a prepaid balance from $10 pays per token.
An abliterated build is the stock model with its refusal direction removed from the weights. It answers prompts the stock model declines, and there is no jailbreak prompt to keep alive. An uncensored build is a fine-tune trained to answer without the stock model’s safety refusals.
Yes. Every build here speaks the OpenAI chat format: point the client at https://api.pentestllm.ai/v1, paste a key and type the model id from the table.
No. An Ollama or GGUF build runs on your own machine and needs the memory for it; these run on the API and are billed per token, so there is nothing to download.