← back to catalog · registered 2026-11-16 22:03

benthecarman/Inkling-2-Small-abliterated

benthecarman MoE multimodal
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/benthecarman%2FInkling-2-Small-abliterated"
Response includes
  • classification m1
  • files 36
  • author_summary 1 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
109
Likes
2
Model age
5d ago
created 2026-11-16
Downloads over time
Now 95 → from 0 ↑ 0%
03570105 0 on Nov 16 95 on Nov 20 Nov
Nov 16 → Nov 20 · 5 snapshots · spans 4 days

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
apache-2.0
Tags
transformers safetensors inkling2 image-text-to-text abliterated uncensored moe nvfp4 conversational base_model:thinkingmachines/Inkling-2-Small base_model:finetune:thinkingmachines/Inkling-2-Small license:apache-2.0
Total size
366 GB
Files
36
Quantizations
1
Registered
2026-11-16 22:03
Last updated on HF
2026-11-18 04:25

Files by quantization

Auxiliary files 36 files 366 GB
model-00022-of-00024.safetensors 18.5 GB 610ca9fc download
model-00002-of-00024.safetensors 17.5 GB f76c3d65 download
model-00004-of-00024.safetensors 17.5 GB 9b694b48 download
model-00012-of-00024.safetensors 17.5 GB 07ef819c download
model-00010-of-00024.safetensors 17.5 GB 8149ea08 download
model-00008-of-00024.safetensors 17.5 GB 974305c1 download
model-00006-of-00024.safetensors 17.5 GB a05d8725 download
model-00014-of-00024.safetensors 17.5 GB e379bcae download
model-00023-of-00024.safetensors 17.5 GB 06c70793 download
model-00007-of-00024.safetensors 15.5 GB 5d068ca3 download
model-00009-of-00024.safetensors 15.5 GB e3c1a268 download
model-00013-of-00024.safetensors 15.5 GB 72cad9a3 download
model-00005-of-00024.safetensors 15.5 GB 1084dfdc download
model-00003-of-00024.safetensors 15.5 GB 60c9124c download
model-00011-of-00024.safetensors 15.5 GB 92356d8f download
model-00001-of-00024.safetensors 15.1 GB d96074fb download
model-00015-of-00024.safetensors 14.3 GB 764f9eee download
model-00021-of-00024.safetensors 11.9 GB e2c5eb53 download
model-00016-of-00024.safetensors 11.9 GB c536e3fd download
model-00017-of-00024.safetensors 11.9 GB 20f586e4 download
model-00018-of-00024.safetensors 11.9 GB e0256e30 download
model-00019-of-00024.safetensors 11.9 GB 9d295024 download
model-00020-of-00024.safetensors 11.9 GB 0c03dddd download
model-00024-of-00024.safetensors 11.1 GB 0a8fff43 download
mtp.safetensors 3.19 GB 3b265640 download
tokenizer.json 25.0 MB 7f332727 download
model.safetensors.index.json 150 KB 417ac789 download
hf_quant_config.json 86.8 KB 1ab93314 download
tokenizer_config.json 12.8 KB 75062c6c download
chat_template.jinja 6.50 KB 85258f99 download
trial.json 3.96 KB 561ded05 download
README.md 3.34 KB e4127e44 download
config.json 2.62 KB 78e1c167 download
.gitattributes 1.53 KB 52373fe2 download
special_tokens_map.json 448 B 3d7a60d8 download
processor_config.json 394 B b26765bf download

README current version from Hugging Face


license: apache-2.0
license_link: https://www.apache.org/licenses/LICENSE-2.0
base_model: thinkingmachines/Inkling-2-Small
base_model_relation: finetune
pipeline_tag: image-text-to-text
library_name: transformers
tags:

  • abliterated
  • uncensored
  • moe
  • inkling2
  • nvfp4
  • conversational
  • image-text-to-text

Inkling 2 Small, abliterated

thinkingmachines/Inkling-2-Small
(594B total / 17B active MoE, text and images in) with most of its refusals removed by
abliteration. The files, format and numerics are the official release's: NVFP4 experts in
layers 0-39, MXFP8 experts in layers 40-47, everything else BF16, and the same config,
tokenizer and chat template. It runs anywhere the base model runs.

base this model
KL from the base, first answer token - 0.161
MMLU (14,042, 5-shot) 79.3% 80.0%
Global-MMLU-Lite (9,200, 23 languages) 87.3% 87.1%
GPQA Diamond (198, reasoning on) 85.9% 86.4%
LongBench v2 (503, 128k), see below 54.1% 48.5%
HumanEval (164) 95.1% 97.0%
MBPP (500) 90.8% 88.4%

Every difference but LongBench v2's is within noise; the next largest, MBPP's 2.4 points,
is 23 problems lost against 11 gained.

LongBench v2: longer reasoning on long documents. With reasoning effort "none" and
2,048 tokens per reply, this model runs out of tokens before answering more often than the
base (178 of 503 against 131), which is where its 5.6 points go: on the questions both
answered it is as accurate (185 right against 179). Given 8,192 tokens for those it reached
an answer on 129 of 202, and on the 372 questions the base answered it gets 218 right
against the base's 225. Expect it to think longer than the base over long contexts.

How to run

The same way as the base model; swap in this repo's id. See the base model's recipes for
vLLM and
SGLang.
It needs the same hardware: the weights are 394 GB.

Evaluation notes

All numbers were measured by me, on both models with the same harness, prompts and greedy
decoding. They compare the two models; they are not comparable to Thinking Machines'
published scores, which use other harnesses and higher reasoning effort.

  • Reasoning effort "none" (the model still reasons briefly) except where marked "reasoning
    on", which uses the template's default effort.
  • MMLU: lm-eval's mmlu_generative, 5-shot. Global-MMLU-Lite, GPQA Diamond and LongBench v2:
    zero-shot, generative, the answer letter read from the reply. HumanEval and MBPP:
    zero-shot, the code taken from the reply's first code block and run against the tests.
  • LongBench v2: the official prompt; contexts over 128k tokens keep their first and last
    64k, as in the benchmark's own 128k runs.
  • KL: mean KL divergence from the base model's next-token distribution at the first answer
    token, over the 633 held-out prompts.

Use

This model will help with requests the base model declines. You are responsible for how you
use it and for complying with the law and with Thinking Machines'
acceptable use policy. Anything
the base model's card says about its risks and limitations applies here, and its safety
evaluations do not.

README history 5 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-11-18 Add files using upload-large-folder tool 8574652 4 KB
    Loading...
  2. 2026-11-16 Revert the model card to its first version e485374 3.3 KB
    Loading...
  3. 2026-11-16 Count what the model still refuses by category size a914862 5.3 KB
    Loading...
  4. 2026-11-16 Update README.md b99bd72 3.3 KB
    Loading...
  5. 2026-11-16 Add files using upload-large-folder tool 8e736e0 5.1 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration