Uncensored LLMs on one API key

7 loosely aligned and refusal-stripped models through a single OpenAI-compatible endpoint, priced per token, with a hard spend ceiling on every key.

What uncensored means here, precisely

The word gets used for two different things, and the difference decides whether a model will actually finish your answer.

An uncensored model is a finetune whose alignment was loosened during training. It refuses far less than a frontier model, and it can still refuse. An abliterated model had the refusal direction removed from its weights outright, so it very nearly cannot. MagmaRouter tags the two separately rather than blurring them, because a page that calls everything uncensored is not telling you which one you are buying.

What this is not It is not a jailbreak, and it is not a way around the law. These are open weights published by their authors, served per token instead of on a GPU you rent. The output is yours and so is the responsibility for it.

The catalogue

Model	Context	In	Out	Tools
DeepSeek V4.1 Flash 1017 Xploded deepseek-v4.1-flash-1017-xploded	1.05M	$1.25	$2.50	Yes
MiMo V2.6 Flash Xploded mimo-v2.6-flash-xploded	1.05M	$1.00	$2.00	Yes
DeepSeek Flash Lite Xploded deepseek-v4.1-flash-lite-1017-xploded	512k	$0.36	$1.44	Yes
Qwen3.8 27B Ultrafast Xploded qwen3.8-27b-fast-xploded	524k	$1.75	$5.50	Yes
Qwen3.8 27B Xploded qwen3.8-27b-xploded	1.05M	$1.50	$5.00	Yes
GLM 5.3 Flash Xploded glm-5.3-flash-xploded	1.05M	$0.30	$1.00	Yes
GLM 5.3 Xploded glm-5.3-xploded	1.05M	$2.00	$5.00	Yes

One base URL

Anything speaking the OpenAI chat completions format works without a code change.

curl https://api.magmarouter.com/v1/chat/completions \
  -H "Authorization: Bearer sk-mr-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.3-flash-xploded",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

FAQ

What counts as an uncensored model?

A finetune whose alignment was loosened by training, so it answers a great deal that a frontier model declines. It is not a guarantee: an uncensored model can still refuse, because the refusal behaviour was reduced rather than removed. The stronger claim is abliteration.

Do I need my own GPU?

No. These are the same open weights you could run locally, served per token. A 70B build at usable quality wants roughly 48GB of VRAM to self-host, which is a card you buy or a GPU you rent by the hour whether or not you use it.

Will it write anything at all?

No, and anyone promising that is selling something. A loosely aligned model answers far more than a frontier one, and it is also more confidently wrong. Removing a refusal does not add knowledge, and it does not change what you are allowed to do with the answer.

Can I cap what a key spends?

Yes. The cap is enforced before the model runs, so the reply cannot cost more than the hold taken before it was sent. That is the difference between a budget and a limit.

Read further

By model family

Uncensored and abliterated builds, family by family, and what to use where MagmaRouter does not sell one.

Looking for a specific model?

The open models people search for uncensored, and what to use for each.

Claude Opus 5.5 uncensoredAnthropic, scores 58 Claude Sonnet 5.5 uncensoredAnthropic, scores 56 Gemini 4 Argon uncensoredGoogle, scores 53 GPT-6 Astra uncensoredOpenAI, scores 53 Claude Fable 5.1 uncensoredAnthropic, scores 53 GPT-6.1 Sol uncensoredOpenAI, scores 52 GPT-6 Sol uncensoredOpenAI, scores 48 K2 Horizon 375B A23B uncensoredK2, scores 47 Mistral Large 4 uncensoredMistral, scores 38 Muse Spark 1.3 uncensoredMeta, scores 48 Kimi K3.1 Pro uncensoredMoonshot AI, scores 55 MiMo V2.6 Pro uncensoredXiaomi, scores 46 Grok 4.7 uncensoredxAI, scores 46 GLM 5.3 uncensoredAvailable uncensored Qwen3.8 Max uncensoredAlibaba, scores 45 Step 5 uncensoredStepFun, scores 44 Kimi K3 uncensoredMoonshot AI, scores 44 GPT-5.6 Terra uncensoredOpenAI, scores 42 GLM 5.3 Flash uncensoredAvailable uncensored Ling 3.1 Flash uncensoredInclusionAI, scores 41 Gemini 3.8 Flash uncensoredGoogle, scores 41 Qwen3.8 2.4T A95B uncensoredAlibaba, scores 40 Qwen3.8 Flash Next uncensoredAlibaba, scores 40 DeepSeek V4.1 Flash uncensoredAvailable uncensored MiMo V2.6 Flash uncensoredAvailable uncensored GPT-6 Luna uncensoredOpenAI, scores 38 DeepSeek V4 Pro uncensoredDeepSeek, scores 36 DeepSeek V4 Flash Vision uncensoredDeepSeek, scores 35 Qwen3.8 27B uncensoredAvailable uncensored K2 Horizon uncensoredK2, scores 31 Gemini 3.1 Pro uncensoredGoogle, scores 30 MiniMax M3 uncensoredMiniMax, scores 29 GPT-5.5 Instant uncensoredOpenAI, scores 26 Kimi K2.7 Code uncensoredMoonshot AI, scores 26 Inkling Small uncensoredThinking Machines, scores 26 Inkling uncensoredThinking Machines, scores 25 Hy3 uncensoredTencent, scores 25 Qwen3.7 Plus uncensoredAlibaba, scores 25 K2 Horizon MoVA 36B uncensoredK2, scores 25 Solar Mini 4 uncensoredUpstage, scores 24 Nemotron 3 Ultra uncensoredNVIDIA, scores 23 Gemini 3.5 Flash-Lite uncensoredGoogle, scores 22 Ling 3.0 Flash uncensoredInclusionAI, scores 20 LongCat 2.0 uncensoredLongCat, scores 19 Qwen3.5 397B A17B uncensoredAlibaba, scores 18 Qwen3.6 35B A3B uncensoredAlibaba, scores 18 Muse Glimmer uncensoredMeta, scores 17 Ring 2.6 1T uncensoredInclusionAI, scores 17 Claude Haiku 4.5 uncensoredAnthropic, scores 17 Qwen3.5 122B A10B uncensoredAlibaba, scores 16 Gemma 4 31B uncensoredGoogle, scores 15 Mistral Medium 3.5 uncensoredMistral, scores 14 Nemotron 3 Super 120B uncensoredNVIDIA, scores 13 Command A+ uncensoredCohere, scores 13