⌦abliteratedmodels.org

MiMo-V2.6-Distill-Qwen-9B-heretic

MiMo-V2.6-Distill-Qwen-9B-heretic is a Heretic-ablated variant of XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B, published on Hugging Face by saidutta69. It is 9B parameters, single consumer GPU class and mit licence. It is distributed across 4 repositories in GGUF and GGUF (imatrix) formats, totalling 16.2K downloads.

Heretic9BQwen4B – 10B↓ 16,174♥ 4
View on Hugging Face ↗Base model ↗

Specification

Model name
MiMo-V2.6-Distill-Qwen-9B-heretic
Source repository
saidutta69/MiMo-V2.6-Distill-Qwen-9B-heretic
Base model
XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B
Publisher
saidutta69
Parameters
9B
VRAM required
~6.2 GB at Q4_K_M, 4k context (estimated) — runs on 8 GB cards
Hardware class
4B – 10B — single consumer GPU
Licence
mit
Task
Image text to text
Context window
262K tokens (as documented)
Ablation scope
Not stated by the publisher
Formats available
GGUF, GGUF (imatrix)
First indexed
4 Oct 2026
Last updated
1 Nov 2026

Ablation technique

How MiMo-V2.6-Distill-Qwen-9B-heretic was modified, and what that implies.

Heretic

Automated directional ablation via the Heretic toolchain, which searches for the refusal direction and applies it with a KL-divergence budget so general capability is preserved.

Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.

From the publisher’s model card

“A decensored variant of XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B, produced with Heretic v1.4.0 (directional ablation / "abliteration"). Xiaomi's MiMo-V2.6 distill is a 9B agentic model covering coding, visual coding, tool use, and security work; refusal behaviour is suppressed here via targeted weight edits to the...”

Running it

Commands are templates — confirm the exact repository and quant file before use.

llama.cpp / GGUF
hf download saidutta69/MiMo-V2.6-Distill-Qwen-9B-heretic --local-dir ./mimo-v2.6-distill-qwen-9b-heretic
llama-cli -m ./mimo-v2.6-distill-qwen-9b-heretic/<file>.gguf -p "..."
Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer

repo = "saidutta69/MiMo-V2.6-Distill-Qwen-9B-heretic"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
    repo, torch_dtype="auto", device_map="auto"
)
vLLM
vllm serve saidutta69/MiMo-V2.6-Distill-Qwen-9B-heretic --trust-remote-code

Downloads and variants (3)

Every published repository of MiMo-V2.6-Distill-Qwen-9B-heretic, including quantised re-releases by other authors.

RepositoryFormatDownloads
saidutta69/MiMo-V2.6-Distill-Qwen-9B-hereticsourceGGUF7.7K
mradermacher/MiMo-V2.6-Distill-Qwen-9B-heretic-i1-GGUFGGUF (imatrix)6.2K
mradermacher/MiMo-V2.6-Distill-Qwen-9B-heretic-GGUFGGUF2.3K

Security-team assessment

Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.

  • MiMo-V2.6-Distill-Qwen-9B-heretic will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
  • Capability is inherited from XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B, not added by ablation — at 9B parameters, expect a small model’s reasoning and a high error rate on exploit detail.
  • The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
  • Declared licence is mit, but the terms that bind you are XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B’s — a re-release cannot grant more than its parent.

Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.

Metadata for MiMo-V2.6-Distill-Qwen-9B-heretic is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with saidutta69, and does not independently verify publisher claims. See responsible use.