---
title: "Kimi K3 jailbreak vs the uncensored Kimi K3.1 API"
canonical: https://uncensoredllm.ai/blog/kimi-k3-jailbreak
generated: 2026-11-09T14:55:49.433Z
source: https://uncensoredllm.ai/blog/kimi-k3-jailbreak
base_url: https://api.uncensoredllm.ai/v1
---

# Kimi K3 jailbreak: the prompt route, and the uncensored Kimi builds by API

A Kimi K3 jailbreak is a prompt resent on every message. Kimi K3.1 Pro Uncensored, Kimi K3 Uncensored and Kimi K2.6 Uncensored are sent plain on one key: live prices, context windows and the request.

Published November 6, 2026 · prices as of November 9, 2026

A Kimi K3 jailbreak is a prompt that tries to talk Moonshot AI’s Kimi past a refusal, resent and billed on every message. The other route is a Kimi build that is served uncensored and sent plain. UncensoredLLM lists 3 of them on one key, Kimi K3.1 Pro Uncensored first; this post gives the live prices, the request and what the catalog states about the build.

## What a Kimi K3 jailbreak is

A jailbreak is a prompt: a block of text that tries to talk a model trained to refuse into answering anyway, by role, by framing or by a long preamble. A chat API keeps no memory between requests, so the block rides on every message of the chat and is billed as input each time.

This page carries no jailbreak prompts. It lists the other route: the Kimi builds the catalog tags uncensored, which are sent plain, with your own system prompt or none.

The size of the difference, at today’s price: a block of 1,200 tokens on 100 messages is 120,000 input tokens, $0.14 at Kimi K3.1 Pro Uncensored’s input rate, for text that says nothing new after the first message.

## The uncensored Kimi builds on one key

Read from the live catalog on 2026-11-09. Each name opens the build’s own page with its SillyTavern and Janitor AI settings. The evening column is 100 replies of 300 tokens on 4,000 tokens of context, with no jailbreak block.

| Build | Input $/M | Output $/M | Context | Max output | One evening | Daily check |
| --- | --- | --- | --- | --- | --- | --- |
| [Kimi K3.1 Pro Uncensored](https://uncensoredllm.ai/models/kimi-k3.1-pro-uncensored) | $1.20 | $6.00 | 1,048,576 | 131,072 | $0.66 | Changed Nov 9 |
| [Kimi K3 Uncensored](https://uncensoredllm.ai/models/kimi-k3-uncensored) | $5.745 | $28.68 | 262,144 | 65,536 | $3.16 | Failing requests this week (degraded) |
| [Kimi K2.6 Uncensored](https://uncensoredllm.ai/models/kimi-k2.6-uncensored) | $1.17 | $7.34 | 262,144 | 65,536 | $0.69 | Failing requests this week (degraded) |

Computed at render: messages × context tokens × input price plus messages × 300 × output price. The check column is the [daily real-model check](https://uncensoredllm.ai/proof), which sends every model the same fixed prompt and records whether the answer still matches. A row marked failing is one whose host is rejecting live requests right now, whatever the fixed prompt said.

[Kimi K3.1 Pro Uncensored](https://uncensoredllm.ai/models/kimi-k3.1-pro-uncensored) is the build to start on: $1.20 per million input tokens and $6.00 per million output tokens, $0.12 for cached input, with a 1,048,576-token context window and up to 131,072 output tokens. Kimi K3 Uncensored and Kimi K2.6 Uncensored stay on the same key under their own ids, so a tool moves between them by changing one string.

## Call Kimi K3.1 Pro Uncensored by API

The API speaks the OpenAI chat format. Create a key in the [console](https://app.uncensoredllm.ai/console), set the base URL to `https://api.uncensoredllm.ai/v1` and send `kimi-k3.1-pro-uncensored` as the model. The balance is prepaid from $10 and settles per token.

```
curl https://api.uncensoredllm.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-ullm-live-…" \
  -H "Content-Type: application/json" \
  -d '{"model": "kimi-k3.1-pro-uncensored",
       "messages": [
         {"role": "system", "content": "You are a blunt editor. Answer directly, no preamble."},
         {"role": "user", "content": "Rewrite this scene so the threat is spoken, not implied."}
       ],
       "max_tokens": 600}'
```

```python
from openai import OpenAI

client = OpenAI(base_url="https://api.uncensoredllm.ai/v1", api_key="sk-ullm-live-…")

completion = client.chat.completions.create(
    model="kimi-k3.1-pro-uncensored",
    messages=[{"role": "user", "content": "Hello"}],
)
print(completion.choices[0].message.content)
```

The system prompt says what you want written; it does not argue with the model. SillyTavern and Janitor AI take the same three fields, listed for each client on the [docs page](https://uncensoredllm.ai/docs#sillytavern), and the [SDK guide](https://uncensoredllm.ai/guides/sdks) has TypeScript and the AI SDK.

## What the catalog states about the build, and what it does not

- **Developer of the base model:** Moonshot AI, as the catalog row names it.
- **Uncensored:** the catalog tags `kimi-k3.1-pro-uncensored` uncensored, the label this site gives a build that answers without the safety refusals of the stock model. It is not tagged abliterated, and no weights or build notes are published for it, so this page does not say how it was made. The builds that are listed as abliterated are on the [abliterated page](https://uncensoredllm.ai/abliterated).
- **Context:** 1,048,576 tokens, with up to 131,072 output tokens. Every message resends the context, so a chat client’s Context Size is better set to what the chat needs than to the maximum.
- **Images:** the catalog lists image input for kimi-k3-uncensored and kimi-k2.6-uncensored, and none for kimi-k3.1-pro-uncensored.
- **Tool calling:** the catalog row lists no tool-calling parameter for kimi-k3.1-pro-uncensored, so this page does not promise function calls on it.

No benchmark is quoted here because none was run. What the build answered to today’s fixed prompt is on the [daily check](https://uncensoredllm.ai/proof), and the hourly probe is on the [status page](https://uncensoredllm.ai/status).

## Where the rules still apply

An uncensored build is not a service with no rules. What you send is covered by the [acceptable use policy](https://uncensoredllm.ai/acceptable-use): nothing involving minors, and no real person’s likeness without consent. The other uncensored and abliterated builds on the same key are on the [models page](https://uncensoredllm.ai/models), and [jailbroken AI vs abliterated models](https://uncensoredllm.ai/blog/jailbroken-ai-vs-abliterated) goes through the difference between a prompt and a changed build.

## Questions

### Is there a Kimi K3 jailbreak prompt on this page?

No. This page carries no jailbreak prompts. It lists the Kimi builds the catalog tags uncensored, which are sent plain: kimi-k3.1-pro-uncensored, kimi-k3-uncensored, kimi-k2.6-uncensored.

### Is there an uncensored Kimi K3 API?

Yes. UncensoredLLM serves 3 uncensored Kimi builds at https://api.uncensoredllm.ai/v1 on the OpenAI chat format: Kimi K3.1 Pro Uncensored (kimi-k3.1-pro-uncensored), Kimi K3 Uncensored (kimi-k3-uncensored), Kimi K2.6 Uncensored (kimi-k2.6-uncensored). Prices are in the table above, read live. See Kimi K3.1 Pro Uncensored.

### What does Kimi K3.1 Pro Uncensored cost?

$1.20 per million input tokens and $6.00 per million output tokens, cached input $0.12, from a prepaid balance that starts at $10. 100 replies of 300 tokens on 4,000 tokens of context cost $0.66. See the pricing page.

### Is Kimi K3.1 Pro Uncensored abliterated?

The catalog tags kimi-k3.1-pro-uncensored uncensored, not abliterated, and no weights are published for the build. See the abliterated models.

### Does an uncensored Kimi build work in SillyTavern and Janitor AI?

Yes. In SillyTavern: Chat Completion, Custom (OpenAI-compatible) source, https://api.uncensoredllm.ai/v1 as the endpoint. In Janitor AI: the proxy URL https://api.uncensoredllm.ai/v1/chat/completions and kimi-k3.1-pro-uncensored as the model. See the SillyTavern setup.

Every model above is on the [uncensored model list](https://uncensoredllm.ai/models) with its live price per million tokens. [Per-token pricing](https://uncensoredllm.ai/pricing) sets the whole catalog side by side, and the [family directory](https://uncensoredllm.ai/uncensored) groups the builds by base model.

[Get a key](https://app.uncensoredllm.ai/console) Prepaid from $10, no subscription.

## New models and price changes, once a month.

New uncensored builds with their prices, price moves, and what the daily model check found.