support@huihui.ai
-
deepseek-r1-abliterated
DeepSeek's first generation reasoning models with comparable performance to OpenAI-o1.
thinking 1.5b 7b 8b 14b 32b 70b518.1K Pulls 55 Tags Updated 2 months ago
-
qwq-abliterated
QwQ is an experimental research model focused on advancing AI reasoning capabilities.
tools 32b63.5K Pulls 17 Tags Updated 4 months ago
-
llama3.2-abliterate
Meta's Llama 3.2 goes small with 1B and 3B models.
tools 1b 3b32.6K Pulls 11 Tags Updated 8 months ago
-
qwen3-abliterated
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.
tools thinking 0.6b 1.7b 4b 8b 14b 16b 30b 32b 235b32.3K Pulls 66 Tags Updated yesterday
-
gemma3-abliterated
The current, most capable model that runs on a single GPU.
1b 4b 12b 27b24.6K Pulls 14 Tags Updated 4 months ago
-
qwen2.5-1m-abliterated
Qwen2.5-1M is the long-context version of the Qwen2.5 series models, supporting a context length of up to 1M tokens.
tools 7b 14b24.2K Pulls 11 Tags Updated 6 months ago
-
qwen2.5-abliterate
Qwen2 is a new series of large language models from Alibaba group
tools 0.5b 1.5b 3b 7b 14b 32b 72b18K Pulls 46 Tags Updated 3 months ago
-
llama3.3-abliterated
New state of the art 70B model. Llama 3.3 70B offers similar performance compared to Llama 3.1 405B model.
tools 70b7,947 Pulls 10 Tags Updated 7 months ago
-
qwen2.5-coder-abliterate
The latest series of Code-Specific Qwen models, with significant improvements in code generation, code reasoning, and code fixing.
tools 0.5b 1.5b 3b 7b 14b 32b7,712 Pulls 31 Tags Updated 8 months ago
-
deepseek-v3
A strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.
7,074 Pulls 2 Tags Updated 6 months ago
-
dolphin3-abliterated
Dolphin 3.0 Llama 3.1 8B 🐬 is the next generation of the Dolphin series of instruct-tuned models designed to be the ultimate general purpose local model, enabling coding, math, agentic, function calling, and general use cases.
tools 8b6,129 Pulls 5 Tags Updated 6 months ago
-
phi4-abliterated
Phi-4 is a 14B parameter, state-of-the-art open model from Microsoft.
14b5,845 Pulls 5 Tags Updated 6 months ago
-
mistral-small-abliterated
Mistral Small 3 sets a new benchmark in the “small” Large Language Models category below 70B.
tools 24b4,145 Pulls 10 Tags Updated 6 months ago
-
huihui-moe-abliterated
Huihui-MoE-abliterated is a **Mixture of Experts (MoE)** language model developed by **huihui.ai**
tools thinking 1.5b 5b 12b 23b 24b 46b 57b2,601 Pulls 36 Tags Updated 1 month ago
-
gemma3n-abliterated
This is an uncensored version of google/gemma-3n created with abliteration
2,460 Pulls 2 Tags Updated 3 weeks ago
-
granite3.2-vision-abliterated
A compact and efficient vision-language model, specifically designed for visual document understanding, enabling automated content extraction from tables, charts, infographics, plots, diagrams, and more.
vision tools 2b2,132 Pulls 5 Tags Updated 5 months ago
-
deepseekr1-qwq-skyt1-fusion
`DeepSeekR1-QwQ-SkyT1-32B-Fusion` is a mixed model that combines the strengths of three powerful Qwen-based models: huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated, huihui-ai/QwQ-32B-Preview-abliterated and huihui-ai/Sky-T1-32B-Preview-abliterated
32b1,896 Pulls 14 Tags Updated 5 months ago
-
phi4-mini-abliterated
Phi-4-mini brings significant enhancements in multilingual support, reasoning, and mathematics, and now, the long-awaited function calling feature is finally supported.
tools 3.8b1,764 Pulls 5 Tags Updated 5 months ago
-
magistral-abliterated
This is an uncensored version of mistralai/magistral-Small-2506 created with abliteration
thinking 24b1,757 Pulls 6 Tags Updated 1 month ago
-
aya-expanse-abliterated
Cohere For AI's language models trained to perform well across 23 different languages.
tools 8b 32b1,596 Pulls 9 Tags Updated 7 months ago
-
skywork-o1-abliterated
The Skywork o1 Open model series, developed by the Skywork team at Kunlun Inc. This groundbreaking release introduces a series of models that incorporate o1-like slow thinking and reasoning capabilities.
tools 8b1,544 Pulls 5 Tags Updated 8 months ago
-
exaone3.5-abliterated
EXAONE 3.5 is a collection of instruction-tuned bilingual (English and Korean) generative models ranging from 2.4B to 32B parameters, developed and released by LG AI Research.
2.4b 7.8b 32b1,498 Pulls 13 Tags Updated 7 months ago
-
qwenlong-abliterated
This is an uncensored version of Tongyi-Zhiwen/QwenLong-L1-32B created with abliteration
thinking 32b1,413 Pulls 5 Tags Updated 2 months ago
-
deephermes3-abliterated
DeepHermes 3 Preview is the latest version of our flagship Hermes series of LLMs by Nous Research, and one of the first models in the world to unify Reasoning (long chains of thought that improve answer accuracy) and normal LLM response modes into one mod
8b1,371 Pulls 6 Tags Updated 5 months ago
-
dolphin3-r1-abliterated
Dolphin's first generation reasoning models.
24b1,284 Pulls 10 Tags Updated 5 months ago
-
granite3.2-abliterated
Granite-3.2 is a family of long-context AI models from IBM Granite fine-tuned for thinking capabilities.
tools 2b 8b1,280 Pulls 11 Tags Updated 5 months ago
-
openthinker-abliterated
A fully open-source family of reasoning models built using a dataset derived by distilling DeepSeek-R1.
7b 32b1,259 Pulls 9 Tags Updated 5 months ago
-
deepseek-v3-pruned
DeepSeek-V3-Pruned-Coder-411B is a pruned version of the DeepSeek-V3 reduced from 256 experts to 160 experts, The pruned model is mainly used for code generation.
411b1,228 Pulls 5 Tags Updated 4 months ago
-
deepseek-v3-abliterated
A strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token.
671b1,207 Pulls 5 Tags Updated 4 months ago
-
marco-o1-abliterated
An open large reasoning model for real-world solutions by the Alibaba International Digital Commerce Group (AIDC-AI).
7b1,159 Pulls 5 Tags Updated 8 months ago
-
tinyr1-abliterated
Tiny-R1-32B-Preview, which outperforms the 70B model Deepseek-R1-Distill-Llama-70B and nearly matches the full R1 model in math.
tools 32b1,156 Pulls 6 Tags Updated 5 months ago
-
acereason-nemotron-abliterated
This is an uncensored version of nvidia/AceReason-Nemotron created with abliteration Edit
7b 14b1,083 Pulls 9 Tags Updated 2 months ago
-
falcon3-abliterated
A family of efficient AI models under 10B parameters performant in science, math, and coding through innovative training techniques.
1b 3b 7b 10b1,056 Pulls 21 Tags Updated 7 months ago
-
arcee-blitz-abliterated
Arcee-Blitz (24B) is a new Mistral-based 24B model distilled from DeepSeek, designed to be both fast and efficient. We view it as a practical “workhorse” model that can tackle a range of tasks without the overhead of larger architectures.
24b962 Pulls 5 Tags Updated 5 months ago
-
kimi-k2
This is not the ablation version. Kimi-K2-Instruct is a state-of-the-art mixture-of-experts (MoE) language model with 32 billion activated parameters and 1 trillion total parameters.
tools 1026b946 Pulls 4 Tags Updated 2 weeks ago
-
tess-r1-abliterated
Tess-R1 is designed with test-time compute in mind, and has the capabilities to produce a Chain-of-Thought (CoT) reasoning before producing the final output.
70b938 Pulls 9 Tags Updated 5 months ago
-
llama3.3-abliterated-ft
The fine-tuned version of huihui_ai/llama3.3-abliterated
tools 70b903 Pulls 4 Tags Updated 7 months ago
-
Hermes-3-Llama-3.2-abliterated
Hermes 3 3B is a small but mighty new addition to the Hermes series of LLMs by Nous Research, and is Nous's first fine-tune in this parameter class.
tools 3b891 Pulls 5 Tags Updated 7 months ago
-
foundation-sec-abliterated
Foundation-Sec-8B-abliterated is an uncensored version fine-tuned based on fdtn-ai/Foundation-Sec-8B. Foundation-Sec-8B is an open-weight, 8-billion-parameter foundational language model designed specifically for cybersecurity applications.
8b884 Pulls 5 Tags Updated 2 months ago
-
am-thinking-abliterate
This is an uncensored version of a-m-team/AM-Thinking-v1 created with abliteration
tools 32b869 Pulls 6 Tags Updated 2 months ago
-
jan-nano-abliterated
This is an uncensored version of Menlo/Jan-nano created with abliteration
tools thinking 4b839 Pulls 10 Tags Updated 1 month ago
-
devstral-abliterated
This is an uncensored version of mistralai/Devstral-Small-2505 created with abliteration
tools 24b839 Pulls 6 Tags Updated 2 months ago
-
internlm3-abliterated
InternLM3 has open-sourced an 8-billion parameter instruction model, InternLM3-8B-Instruct, designed for general-purpose usage and advanced reasoning.
8b814 Pulls 5 Tags Updated 6 months ago
-
kanana-nano-abliterated
Kanana, a series of bilingual language models (developed by Kakao) that demonstrate exceeding performance in Korean and competitive performance in English.
2.1b812 Pulls 5 Tags Updated 5 months ago
-
qwq-fusion
qwq-fusion is a mixed model that combines the strengths of two powerful Qwen-based models: huihui-ai/QwQ-32B-Preview-abliterated and huihui-ai/Qwen2.5-Coder-32B-Instruct-abliterated.
tools 32b800 Pulls 14 Tags Updated 8 months ago
-
exaone-deep-abliterated
This is an uncensored version of LGAI-EXAONE/EXAONE-Deep-2.4B created with abliteration
2.4b 7.8b751 Pulls 9 Tags Updated 1 month ago
-
granite3.1-dense-abliterated
The IBM Granite 2B and 8B models are text-only dense LLMs trained on over 12 trillion tokens of data, demonstrated significant improvements over their predecessors in performance and speed in IBM’s initial testing.
tools 2b 8b730 Pulls 11 Tags Updated 7 months ago
-
baronllm-abliterated
This is an uncensored version of AlicanKiraz0/BaronLLM_Offensive_Security_LLM_Q6_K_GGUF created with abliteration
tools 8b729 Pulls 4 Tags Updated 1 month ago
-
perplexity-ai-r1-abliterated
A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.
70b716 Pulls 10 Tags Updated 5 months ago
-
deepscaler-abliterated
A fine-tuned version of Deepseek-R1-Distilled-Qwen-1.5B that surpasses the performance of OpenAI’s o1-preview with just 1.5B parameters on popular math evaluations.
1.5b670 Pulls 3 Tags Updated 5 months ago
-
homunculus-abliterated
This is an uncensored version of arcee-ai/Homunculus created with abliteration
thinking 12b635 Pulls 5 Tags Updated 1 month ago
-
command-r7b-abliterated
he smallest model in Cohere's R series delivers top-tier speed, efficiency, and quality to build powerful AI applications on commodity GPUs and edge devices.
tools 7b593 Pulls 5 Tags Updated 6 months ago
-
phi4-reasoning-abliterated
Phi 4 mini reasoning is a lightweight open model that balances efficiency with advanced reasoning ability.
3.8b587 Pulls 4 Tags Updated 3 months ago
-
smallthinker-abliterated
A new small reasoning model fine-tuned from the Qwen 2.5 3B Instruct model.
3b569 Pulls 5 Tags Updated 7 months ago
-
deepseek-r1-Fusion
DeepSeek-R1-Distill-Qwen-Coder-32B-Fusion-9010 is a mixed model that combines the strengths of two powerful DeepSeek-R1-Distill-Qwen-based models: huihui-ai/DeepSeek-R1-Distill-Qwen-32B-abliterated and huihui-ai/Qwen2.5-Coder-32B-Instruct-abliterated.
32b564 Pulls 6 Tags Updated 5 months ago
-
tulu3-abliterate
Tülu 3 is a leading instruction following model family, offering fully open-source data, code, and recipes by the The Allen Institute for AI.
8b 70b547 Pulls 9 Tags Updated 8 months ago
-
microthinker
MicroThinker is an experimental research model focused on advancing AI reasoning capabilities.
tools 1b 3b 8b495 Pulls 12 Tags Updated 6 months ago
-
nemotron-v1-abliterated
Llama Nemotron, Open, Production-ready Enterprise Models
tools 8b481 Pulls 6 Tags Updated 3 months ago
-
seed-coder-abliterate
This is an uncensored version of ByteDance-Seed/Seed-Coder-8B-Instruct created with abliteration
469 Pulls 5 Tags Updated 2 months ago
-
openhands-lm-abliterated
OpenHands LM is built on the foundation of Qwen Coder 2.5 Instruct 32B, leveraging its powerful base capabilities for coding tasks.
tools 32b410 Pulls 7 Tags Updated 4 months ago
-
perplexity-ai-r1
A version of the DeepSeek-R1 model that has been post trained to provide unbiased, accurate, and factual information by Perplexity.
372 Pulls 2 Tags Updated 5 months ago
-
Qwen3-Coder
This is not the ablation version. Qwen3-Coder featuring the following key enhancements: Significant Performance, Long-context Capabilities, Agentic Coding.
tools thinking 480b362 Pulls 4 Tags Updated 1 week ago
-
nemotron-abliterated
Llama-3.1-Nemotron-70B-Instruct is a large language model customized by NVIDIA to improve the helpfulness of LLM generated responses to user queries.
tools 70b352 Pulls 6 Tags Updated 7 months ago
-
s1.1-abliterated
This model is a successor of s1-32B with slightly better performance.
tools 32b339 Pulls 4 Tags Updated 5 months ago
-
s1-abliterated
s1 is a reasoning model finetuned from Qwen2.5-32B-Instruct on just 1,000 examples. It matches o1-preview & exhibits test-time scaling via budget forcing.
tools 32b306 Pulls 9 Tags Updated 5 months ago
-
huihui-moe
Huihui-MoE is a Mixture of Experts (MoE) language model developed by huihui.ai
tools thinking 1.2b 23b285 Pulls 10 Tags Updated 1 month ago
-
uwu-abliterated
UwU is an experimental research model focused on advancing AI reasoning capabilities.
7b265 Pulls 5 Tags Updated 7 months ago
-
Lucie-abliterated
Lucie-7B is a pretrained 7B parameter causal language model built by LINAGORA and OpenLLM-France.
7b231 Pulls 5 Tags Updated 6 months ago
-
qwen3-coder-abliterated
Qwen3-Coder featuring the following key enhancements: Significant Performance, Long-context Capabilities, Agentic Coding.
30b205 Pulls 6 Tags Updated 2 days ago
-
fluentlylm-prinum-abliterated
fluently-lm/FluentlyLM-Prinum
tools 32b184 Pulls 5 Tags Updated 5 months ago
-
skyt1-abliterated
Sky-T1-32B-Preview is an experimental research model focused on advancing AI reasoning capabilities
tools 32b177 Pulls 5 Tags Updated 6 months ago
-
deepseek-r1
DeepSeek's first-generation of reasoning models with comparable performance to OpenAI-o1, including six dense models distilled from DeepSeek-R1 based on Llama and Qwen.
thinking 671b144 Pulls 12 Tags Updated 1 month ago
-
megrez-abliterated
Megrez-3B aims to provide a fast inference, compact, and powerful edge-side intelligent solution through software-hardware co-design.
tools 7b130 Pulls 6 Tags Updated 6 months ago
-
qwen2.5-censortune
CensorTune with Supervised Fine-Tuning (SFT) to fine-tune the Qwen2.5-Instruct model on 622 harmful instructions in a single fine-tuning iteration, achieving rejection of these instructions and a zero-pass rate for 320
tools 0.5b 1.5b 3b87 Pulls 15 Tags Updated 3 months ago
-
deepseek-r1-pruned
DeepSeek-R1-Pruned-Coder-411B is a pruned version of the DeepSeek-R1 reduced from 256 experts to 160 experts, The pruned model is mainly used for code generation.
411b75 Pulls 3 Tags Updated 4 months ago