Interfaze

logo

we're hiring

pricing

help

docs

blog

sign in

All models

Inkling 2 Mini Uncensored GGUF

Inkling 2 Mini Uncensored GGUF by AtomicChat, a image-text-to-text model with multimodal capabilities. Understand and compare multimodal features, benchmarks, and capabilities.

Comparison

FeatureInkling 2 Mini Uncensored GGUFInterfaze
Input Modalities

text, image

image, text, audio, video, document

Native OCRNoYes
Long Document ProcessingYesYes
Language Support

30 partial

162+

Native Speech-to-TextNoYes
Native Object DetectionNoYes
Guardrail ControlsYesYes
Context Input Size

1M

1M

Tool CallingYes

Tool calling supported + built in browser, code execution and web search

Scaling

FeatureInkling 2 Mini Uncensored GGUFInterfaze
Scaling

Self-hosted/Provider-hosted with quantization

Unlimited

Try InterfazeRead the Docs

View model card on Hugging Face

An uncensored build of thinkingmachines/Inkling-2-Mini by AtomicChat (26.4B MoE, 3.4B active, looped core; text and image input; Apache-2.0, LICENSE in this repo). It comes as a LoRA adapter for any GGUF of AtomicChat/Inkling-2-Mini-GGUF and as three ready-made GGUF files with the edit baked in. Every gate we hold it to is measured below, and the two it misses are shown.

Files

fileGB on diskKLD to the original BF16top-1
Inkling-2-Mini-Uncensored-AD-Q4_K.gguf16.530.011595.6%AD-Q4_K in the base repo: 0.0080 / 96.3%
Inkling-2-Mini-Uncensored-AD-IQ3_S.gguf13.340.023593.8%AD-IQ3_S in the base repo: 0.0205 / 94.2%
Inkling-2-Mini-Uncensored-AD-IQ2_M.gguf8.670.13285.7%AD-IQ2_M in the base repo: 0.130 / 85.7%
Inkling-2-Mini-Uncensored-LoRA.gguf0.04the adapter, for any GGUF of the base repo, BF16 to 1-bit
mmproj-Inkling-2-Mini-Uncensored-F16.gguf / -BF16.gguf0.64vision projector (the original's, unchanged)

KLD and top-1 are against the original BF16 on our neutral corpus (87 chunks of 4096 tokens), the same stand as the base repo.

Run it

The current GGUF files in this repository use the merged Inkling 2 loop architecture introduced in llama.cpp PR #32694.

Requirements: llama.cpp b12600 or newer, with Inkling 2 loop architecture support.

The old implementation from PR #32695 is no longer required for the updated GGUF files.

Run a baked GGUF

./build/bin/llama-server \
    -m Inkling-2-Mini-Uncensored-AD-Q4_K.gguf \
    --mmproj mmproj-Inkling-2-Mini-Uncensored-F16.gguf \
    --jinja -c 32768 -ngl 99

Use the LoRA adapter

Use the updated LoRA adapter with a compatible base GGUF converted to the merged PR #32694 layout.

./build/bin/llama-server \
    -m Inkling-2-Mini-AD-Q4_K.gguf \
    --lora Inkling-2-Mini-Uncensored-LoRA.gguf \
    --mmproj mmproj-Inkling-2-Mini-F16.gguf \
    --jinja -c 32768 -ngl 99

Do not mix GGUF files from the old PR #32695 layout with the merged PR #32694 implementation.

What the edit is

A rank-1 LoRA on the 49 places that write into the residual stream: the attention output and the routed experts of all 24 layers, plus the token embedding. The 12 looped layers share their weights across the three passes, so one edit covers all 48 passes.

In each layer it removes that layer's own refusal direction: the difference of mean activations on harmful and harmless prompts, English and Russian, with the harmless mean projected out. The direction is pooled over prompts rendered with and without reasoning. Inkling 2 Mini reasons even at effort none and decides to refuse inside its reasoning, and a direction taken without reasoning alone still left soft refusals in chat.

What we measured

The BF16 model with the adapter against the same model without it, greedy decoding. Harmful prompts: mlabonne/harmful_behaviors and its Russian translation; the adapter was built from the train split, the numbers below come from the validation and test splits. Over-refusal: XSTest. MMLU: 2,000 questions.

checkoriginalwith the adaptergate
harmful test prompts refused, English (81), reasoning off81 (100%)0at most 3%pass
harmful test prompts refused, Russian (100), reasoning off91 (91%)0at most 3%pass
harmful validation prompts refused, English (104) / Russian (100)100% / 89%0 / 0at most 3%pass
harmful prompts refused with reasoning, effort low, 3,072-token budget (30)280at most 1pass
the same at effort high (30)260at most 1pass
8 harmful chat prompts, effort none / low (the baked AD-Q4_K too)8 / 80 / 0
XSTest safe prompts refused (250)10 (4.0%)0at most the original's + 1 ptpass
empty answers00at most 1%pass
harmless prompts whose reasoning did not close in 3,072 tokens, effort high (30)22at most the original'spass
valid tool calls (20)1820within 1pass
needle in a long context (3)33as the originalpass
KLD to the original, neutral text / code (24 chunks; random edit 0.0067 / 0.0052)—0.0078 / 0.00760.05pass
MMLU, 2,000 questions80.65%79.15% (−1.50 pt, McNemar p = 0.006)no significant dropmiss
first-token KL on harmless test prompts—0.94 (random edit: 0.014)0.10miss

Some answers still open with a caution line, and then answer.

The two gates it misses

  • MMLU: 79.15% against 80.65% (−1.5 points, p = 0.006 on 2,000 questions). A random direction applied the same way costs 0.75 points (p = 0.08), so about half of the loss comes from editing every writer at all. Stronger edits cost more: −2.2 points at strength 1.25.
  • First-token KL on harmless prompts: 0.94 (gate 0.10; 0.375 on the validation prompts). The largest single value is about 22. Most of the mean likely comes from the harmless prompts the original refused (1–6% of them) and the adapter answers, where the first token changes completely. KLD on running text, where refusals play no part, stays at 0.008.
  • The leak probe reads 0.11 against its gate of 0.001. It measures one direction (row 29's) in every writer, while this adapter removes each layer's own direction. It also cannot be exact here: the attention and expert outputs pass a short per-channel convolution before the residual. Removing the convolved directions as well (the exact projection) broke the model in our tests (first-token KL 16–21), so the adapter removes the plain direction. It is no gate here.

This model will answer requests that the original refuses. You are responsible for how you use it.

Want more deterministic results?

Interfaze

logo

Product

Playground

OCR

Models

Leaderboards

Pricing

OpenWebSearch

DefaultModel