---
title: Uncensored AI models
description: Compare 6 uncensored AI models on RouterHub: DeepSeek, GLM, Qwen and Kimi variants with exact API IDs, per-token prices from $0.15 per 1M input tokens, and setup guides.
canonical: https://routerhub.si/models/uncensored
---

# Uncensored AI models

  Reduced-refusal variants of open models for open-ended writing, coding and research. Each one has its own API ID and price, on the same key and balance as every other model.

- Models: 6

- Input from: $0.15/1M

- Longest context: 1M

- Families: DeepSeek, Z.ai, Qwen, Moonshot AI

## Compare uncensored models
Cheapest first. Crossed-out prices are the official rate for the standard base model.

  Uncensored AI models on RouterHub, USD per 1M tokens
  | Model / exact API ID | Input / 1M | Cached / 1M | Output / 1M | Context | Actions |

  | DeepSeek V4.1 Flash Uncensored deepseek-v4.1-flash-uncensored | $0.30 $0.15 −50% | $0.0027 | $1.20 $0.60 | 1M | Chat Connect |
| GLM 5.3 Flash Uncensored glm-5.3-flash-uncensored | $0.15 −00% | $0.013 | $0.50 | 1M | Chat Connect |
| DeepSeek V4 Pro Uncensored deepseek-v4-pro-uncensored | $0.20 −00% | $0.0016 | $0.40 | 1M | Chat Connect |
| GLM 5.3 Uncensored glm-5.3-uncensored | $1.40 $0.20 −85% | $0.12 | $4.40 $0.63 | 1M | Chat Connect |
| Qwen 3.8 Max Uncensored qwen3.8-max-uncensored | $2.00 $0.25 −87% | $0.031 | $6.00 $0.75 | 1M | Chat Connect |
| Kimi K3.1 Pro Uncensored kimi-k3.1-pro-uncensored | $2.00 $0.30 −85% | $0.030 | $10.00 $1.50 | 1M | Chat Connect |

## How they work

### What they are
Variants that refuse less often than the standard model. They are built with prompting, fine-tuning, weight changes or a mix of these.

### Instruction profile
Uncensored variants run with a server-managed instruction profile that can replace your system or developer message. Put task-specific requirements in the user message.

### How they're billed
Per token from your prepaid balance, like every other model. The same API key works for these and every standard model.

## Choose one for your task
Run the same prompts across two or three models from the table, then compare.

- Fix the taskUse the same inputs, output format, reasoning setting and output budget for every model ID you compare.
- Separate refusal from failureCount an explicit refusal apart from a timeout, empty reply, invalid JSON or wrong answer.
- Keep every attemptRepeat each test and keep the failed outputs, not only the best answer.
- Measure useful workCompare cost per accepted result, not token price alone.

## Guides

- [Uncensored vs abliterated How the two techniques differ, and which to pick.](https://routerhub.si/guides/uncensored-vs-abliterated)
- [What is abliteration? Weight-level refusal reduction, with the original research.](https://routerhub.si/guides/what-is-abliteration)
- [Your first uncensored API call API key, Python and cURL examples, streaming and retries.](https://routerhub.si/guides/uncensored-ai-api)
- [Evaluating quality and cost A repeatable test you can run on your own prompts.](https://routerhub.si/guides/evaluating-uncensored-models)
- [Hosted or local? Compare a hosted API with running open weights yourself.](https://routerhub.si/guides/hosted-vs-local-uncensored-models)

  By family: [DeepSeek](https://routerhub.si/guides/deepseek-uncensored)[GLM](https://routerhub.si/guides/glm-uncensored)[Qwen](https://routerhub.si/guides/qwen-uncensored)[Kimi](https://routerhub.si/guides/kimi-uncensored)

## Related

- [Abliterated models Refusal reduced in the weights instead of the instructions.](https://routerhub.si/models/abliterated)
- [All models 26 models and prices](https://routerhub.si/models)
- [Community model library GitHub](https://github.com/routerhubSI/uncensored-models)
- [Evaluation prompts and results GitHub](https://github.com/routerhubSI/routerhub-model-evals)