Best Uncensored Local LLMs: Abliterated and Unfiltered Models That Run on Ollama

Short answer: start with dolphin3:8b — Dolphin 3.0 on a Llama 3.1 base, ~5GB at Q4, 128K context, runs on any 8GB GPU. If you have 12-16GB of VRAM, huihui_ai/qwen3-abliterated:14b (~9GB) is the strongest current-generation uncensored pick, and at 48GB the ceiling is huihui_ai/llama3.3-abliterated:70b. Every model on this page is a real, pullable Ollama tag.

Two things before the list, because most pages on this topic skip both. First, "uncensored" covers two very different techniques — a dataset-filtered fine-tune and a weight-surgery trick called abliteration — and they have different quality costs. Second, those costs are real: abliterated models are measurably worse at reasoning than their stock versions, and we will tell you by how much it matters rather than pretend otherwise. If you just want our general Ollama rankings with filters intact, that is the best Ollama models guide; this page is for when the stock model refuses work you have every right to do — dark fiction, security research, medical or legal questions phrased bluntly, villain dialogue that a corporate safety layer keeps rewriting into an apology.

---

What "Uncensored" Actually Means {#what-uncensored-means}

Two techniques produce "uncensored" models: filtering alignment out of the training data (Dolphin's approach), or abliteration — surgically deleting the refusal direction from a finished model's weights. They are not interchangeable, and knowing which one you are pulling predicts how the model will behave.

Dataset-filtered fine-tunes. Eric Hartford's Dolphin series takes a strong base model and fine-tunes it on instruction data that has been filtered to remove alignment and moralizing responses. The model never learns the refusal reflex in the first place. Because this is a normal fine-tune, quality holds up well — Dolphin models behave like polished assistants that simply do not lecture you.

Abliteration. A finished, aligned model gets its weights edited directly. The insight behind it: refusal behavior in chat models is largely mediated by a single direction in the model's activation space. Find that direction, subtract it out of the weights, and the "I can't help with that" reflex disappears while the knowledge stays. It is fast and cheap — which is why abliterated builds of new models appear on Ollama within days of release — but it is surgery on a live patient. The most prolific publisher, huihui_ai, describes its own method as a crude, proof-of-concept implementation for removing refusals. Respect that honesty; it matches how the models behave.

What neither technique does: add knowledge the base model never had, or make output legal that was not legal before. An uncensored model is the same model with the "no" removed — nothing more.

---

The Dolphin Family: The Uncensored Default {#dolphin-family}

If you want one uncensored model and no research project, pull dolphin3:8b. The Dolphin line has been the reference uncensored fine-tune since the Llama 2 days, and the current generation is genuinely good.

Speed matches the base weights: on an RTX 4060, expect ~45 tok/s from the 8B Dolphins and ~52 tok/s from dolphin-mistral, same as the stock Llama 3.1 8B and Mistral 7B they are built on.

One thing worth quoting from Dolphin's own model card, because it frames the whole category correctly: the model is steerable and gives control to the system owner — you set the system prompt, you decide the alignment — and the authors explicitly advise implementing your own alignment layer before exposing it as a service. In other words: built for you running it on your machine, not for you wiring it into a public chatbot unfiltered.

---

Abliterated Builds of Current Models {#abliterated-builds}

When you want this year's base models without refusals, the huihui_ai namespace on Ollama is where they live — abliterated builds of Qwen 3, Gemma 3, Llama 3.3 and DeepSeek-R1, published as ordinary pullable tags.

At the same quantization, footprints match the stock models — abliteration edits weights, it does not shrink them — so our measured base-model numbers carry over: the 14B runs ~35 tok/s on an RTX 4080, the 8B-class models 44-50 tok/s on an RTX 4060. The catch is that huihui_ai does not publish every line at Q4. The Qwen 3 builds default to Q4_K_M, but gemma3-abliterated only exists at q8_0 and fp16 — so check the file size on the tag listing before you pull, not the stock model's.

The trade: these are the newest, smartest bases you can get uncensored, and also the builds where the quality cost of the surgery shows most. More on that below — read the trade-off section before you make one of these your daily driver.

---

Hermes 3: The Steerable Middle Ground {#hermes-3}

Hermes 3 (ollama pull hermes3:8b) is not marketed as "uncensored" — it is neutrally aligned, which in practice means the system prompt decides everything. Nous Research fine-tuned Llama 3.1 to follow instructions exactly and neutrally, without refusal behavior baked into the weights; the behavioral boundaries are whatever you write in the system prompt.

For a lot of people this is actually the right answer. Nothing was filtered out and nothing was ablated, so Hermes 3 keeps essentially all of its base-model quality — it is a genuinely strong generalist with good roleplay, reasoning and long-context coherence, sitting in the official Ollama library at 3b and 8b (70b and 405b exist for bigger hardware). Write a system prompt that says who the model is and what it will and will not do, and it follows that — yours, not Meta's.

The honest caveat: neutral alignment is not zero alignment. With an empty system prompt Hermes 3 is more conservative than Dolphin or an abliterated build. It rewards people who actually use system prompts; if you never touch them, Dolphin gives you the uncensored default with less setup.

---

The Llama-2-Era Classics — and Why to Skip Them {#legacy-picks}

llama2-uncensored, wizard-vicuna-uncensored and wizardlm-uncensored still sit in the Ollama library, and in 2026 we would not install any of them. These early builds invented the category — Eric Hartford's Wizard pair plus George Sung and Jarrad Hope's llama2-uncensored — and wizard-vicuna-uncensored (7b/13b/30b) was the uncensored model for a generation of local AI users. But but they are built on Llama 2-era bases with short contexts and reasoning that a modern 8B beats without trying.

They show up in every "uncensored Ollama" listicle because those listicles copy each other. The only reason to pull one today is curiosity about where the scene came from. For actual use, dolphin3:8b is smaller than the old 13Bs and better at everything.

---

VRAM: What Fits Your Card {#vram-requirements}

The 8B-class picks need ~5GB, the 14B needs ~9GB, the 70B needs ~40GB — at the same quant, uncensored builds cost the same VRAM as their stock versions (just check the tag, since not every abliterated line is published at Q4). Match the pick to your card:

The usual fitting rules apply — Q4_K_M is the sweet spot, and it is the KV-cache growing with context that triggers out-of-memory errors, not the weights. Our tier guides cover the details for your card: 8GB, 12GB, 16GB and 24GB.

---

The Quality Trade-Off: What You Give Up {#quality-trade-offs}

Abliteration makes models measurably worse — sloppier instruction-following and degraded multi-step reasoning — and anyone selling you an abliterated build as a free lunch is wrong. The refusal direction does not live in clean isolation; subtracting it out of the weights nicks the capabilities stored near it. How much varies by model and by how aggressively the ablation was run, which is why the community keeps re-releasing improved versions of the same abliterations.

The practical hierarchy, from least to most quality lost:

1. Hermes 3 — nothing removed, nothing ablated. Essentially full base-model quality; you pay in system-prompt effort instead.
2. Dolphin — a normal fine-tune on filtered data. Holds up well; the cost is that its bases trail the newest releases by a generation.
3. Abliterated builds — newest bases, real surgery. Fine for conversation and creative work; noticeably less reliable on hard multi-step reasoning and precise formatting than their stock versions.

Our advice from running these side by side: keep a stock model installed next to your uncensored one and switch per task. Uncensored for the writing, roleplay and blunt questions the stock model refuses; stock for the code review and the reasoning-heavy work. Ollama makes holding both a non-issue — a second 5GB model costs you disk, not VRAM, until you load it.

---

Frontends: Where the Creative-Writing Crowd Runs These {#frontends}

Ollama runs every model on this page, but for long-form fiction and roleplay most of this community runs KoboldCpp or text-generation-webui instead — same GGUF weights, better tooling for the job. KoboldCpp adds character cards, world info and memory management that a chat UI does not have; text-generation-webui exposes the sampler controls that matter when you are steering a model through a 20K-token story. We keep setup guides for both: KoboldCpp and text-generation-webui.

And if you came here for the image side of this question — running FLUX and SDXL without content filters — that is its own page: uncensored local image generation.

---

Responsible Use: Local Is Not Lawless {#responsible-use}

Running these models is legal, and the legitimate reasons to run them are real — fiction with actual villains, security research, unfiltered answers to medical and legal questions, and simply not having a corporation's content policy sitting between you and your own hardware. But be clear-eyed about what you have removed: a safety layer, not a law. Whatever is illegal to produce or possess in your jurisdiction remains exactly as illegal when a local model generates it, and you are the publisher of anything you do with the output. There is no place on your machine or anyone else's for content that exploits real people or breaks the law — and the model's own authors say the same thing: Dolphin's card tells you to add your own alignment layer before exposing it as a service, precisely because "uncensored on my desk" and "unfiltered on the public internet" are different decisions. Run them privately, keep responsibility for the output, and this whole category stays what it should be: a tool for adults.

---

Sources {#sources}

• Ollama library — dolphin3, dolphin-llama3, dolphin-mistral, dolphin-mixtral, hermes3, llama2-uncensored, wizard-vicuna-uncensored, wizardlm-uncensored (all tags verified at publish time)
• huihui_ai on Ollama — qwen3-abliterated, gemma3-abliterated, llama3.3-abliterated, deepseek-r1-abliterated, dolphin3-abliterated tags and model cards
• Nous Research — Hermes 3 technical report — neutral-alignment design
• Cognitive Computations (Eric Hartford) — Dolphin model cards and dataset-filtering methodology

---

FAQ {#faq}