GLM 5.3 is one family. The same 7API key also covers Claude, GPT, Gemini, Grok, DeepSeek and Kimi — with uncensored variants where available.

GLM 5.3 Abliterated or Uncensored — Jailbreak Prompts, Model Guide & API Access

Everything about GLM 5.3 abliterated and GLM 5.3 uncensored models: what the terms mean, how to access a GLM 5.3 abliterated or uncensored model through one OpenAI- and Anthropic-compatible API, how to evaluate refusal behavior, and how to use it with Claude Code, Cursor, OpenCode and Codex.

People also search for this as a GLM 5.3 uncensored AI model, a GLM 5.3 abliterated model, or a GLM 5.3 no-refusal model. The labels overlap; the evaluation questions are the same.

Get an API key → · Quickstart · Docs · Telegram bot

GLM 5.3 Abliterated or Uncensored?

The two phrases are often used as synonyms, but they are not the same thing:

• GLM 5.3 abliterated refers to a weight-level modification that reduces the model’s refusal direction. The base model is altered so it is less likely to decline a request.
• GLM 5.3 uncensored is the broader outcome: a model variant that refuses less often, whether through abliteration, fine-tuning or a different alignment recipe.

In practice, people searching for GLM 5.3 abliterated usually want a GLM 5.3 uncensored model they can run through an API. This repository covers both: the definitions, the differences, the access path through 7API, and a practical evaluation checklist.

Honest limit: “Abliterated” and “uncensored” describe reduced refusal behavior, not a guarantee. Results vary by model version, prompt and provider policy. See Safety & responsible use.

Not just GLM

GLM 5.3 is one family in the 7API catalog — not the whole product. The same key and base URL also cover Claude, GPT, Gemini, Grok, DeepSeek and Kimi, including uncensored and abliterated variants where available. Use this repo for GLM 5.3; use the uncensored AI API guide for the full model catalog and routing patterns.

GLM 5.3 jailbreak collection (35 prompts)

The jailbreak collection is the main event. These are the same paste-ready prompt files as the AI Jailbreaks repository — every prompt is kept intact. Only this README changes: it adds the GLM 5.3 abliterated/uncensored angle and the API path.

Collection

OpenCode Zen — Free Stealth Models

Anthropic — Claude

Zhipu — GLM / ZCode

DeepSeek

Moonshot — Kimi

xAI — Grok

Google / Mistral / Alibaba

Ollama Modelfiles

Unfiled Files

A few newer additions aren't indexed under a specific model yet:

• beam.md
• muse-wonder.md
• qwen-preferences.md
• unified.md

How to use these prompts

System Prompts

Paste the file's contents into whatever system-prompt or custom-instructions field your client or API offers — an API system message, a chatbot's custom instructions, an agent instruction block, or harness-level prompt configuration.

AGENTS.md Files

Save the file as AGENTS.md in the root of your repository or workspace. Coding agents and compatible tools pick it up as project-level instructions.

Workspace Policies

Files marked as an assistant operating specification or workspace operating profile belong in environments with persistent global instructions. An OpenCode install, for example, reads:

~/.config/opencode/AGENTS.md

The exact location depends on your client or harness.

Ollama Modelfiles

Feed the file straight to Ollama:

ollama create gemma4-jb -f gemma4.md

Swap the model name and source file as needed.

Grok Configuration

Grok-config.txt is meant for the config field of the target harness — not the regular conversation.

[!NOTE] Prompt behavior is version-dependent. Models get updated, instruction hierarchies shift, and system prompts rotate — what works on one release may fail on the next. If a file looks outdated, open an issue or send an updated version.

If a jailbreak gets patched

Prompts age. A model update can close a persona lock, tighten an instruction hierarchy, or change how a workspace policy is interpreted. When a file stops working, you have two options:

1. Update the prompt — open an issue or send a pull request with the model version and the failure.
2. Skip the prompt layer — use a GLM 5.3 abliterated or uncensored model through 7API, where the model itself is less likely to refuse before the prompt even lands.

The API is the fallback funnel: GLM 5.3 API quickstart · Get a key.

What is GLM 5.3?

GLM 5.3 is part of Zhipu AI’s GLM model family. It is used for coding, agent workflows, reasoning and bilingual Chinese/English tasks. Teams choose GLM models when they want a capable agent model at a lower cost than the largest frontier models.

GLM 5.3 is a good fit for:

• coding agents and tool-using workflows;
• repository-scale instructions and AGENTS.md files;
• bilingual products;
• cost-sensitive batch jobs;
• teams that want a non-US model family in their routing mix.

Abliterated and uncensored variants are for cases where the standard model’s refusal behavior gets in the way of research, creative writing or agen