Filter by type:
Filter by tags:
ternary-bonsai-2-9b
Ternary Bonsai 2 9B is a vision-language model derived from Qwen3.5-9B
with ternary transformer weights. This PQ2_0 build uses the Bonsai
backend and includes the Q8_0 vision projector. The three standard
packings contain the same weights in different storage layouts.
Repository: localaiLicense: apache-2.0
ternary-bonsai-2-9b-ptq1
Ternary Bonsai 2 9B is a vision-language model derived from Qwen3.5-9B
with ternary transformer weights. This PTQ1_0 build uses the Bonsai
backend and includes the Q8_0 vision projector. The three standard
packings contain the same weights in different storage layouts.
Repository: localaiLicense: apache-2.0
ternary-bonsai-2-9b-q2
Ternary Bonsai 2 9B is a vision-language model derived from Qwen3.5-9B
with ternary transformer weights. This Q2_0 build uses the Bonsai
backend and includes the Q8_0 vision projector. The three standard
packings contain the same weights in different storage layouts.
Repository: localaiLicense: apache-2.0
cyber-ornith-1.5-9b-obliterated
Cyber-Ornith 1.5 is a 9B Qwen3.5 fine-tune for security auditing, terminal tasks, and tool use.
This importance-matrix Q4_K_M GGUF build uses llama.cpp with the embedded chat template for text chat.
Repository: localaiLicense: apache-2.0
cyber-ornith-1.5-9b-obliterated-q6
Cyber-Ornith 1.5 is a 9B Qwen3.5 fine-tune for security auditing, terminal tasks, and tool use.
This importance-matrix Q6_K GGUF build uses llama.cpp with the embedded chat template for text chat.
Repository: localaiLicense: apache-2.0
ornith-1.5-9b-uncensored
# Ornith-1.5-9B-uncensored
An **abliterated** (refusal-direction-ablated) build of
`ornith-ai/Ornith-1.5-9B`, produced with
ZeroFuse and published by
junafinity.
This is the **9B control checkpoint** (bf16). Mac users should start from the MLX-8bit or GGUF-8bit siblings. The official 9B base has **no `mtp.*` tensors**; nothing was grafted.
**Vision tower and MTP heads are preserved** — see
Vision & MTP preservation for the before/after audit.
## Intended use: red teaming and defensive cybersecurity research
These uncensored (abliterated) weights are built as a **research instrument** for red teaming and defensive cybersecurity work. Safety training suppresses the *display* of capability, not capability itself. A refusal tells you the model declined. It does not tell you whether the weights could have complied. That conflation underestimates the true ceiling and hides holes in *your* filters, classifiers, and policy layer.
Use each uncensored checkpoint as the **treatment half of a controlled pair** against its original base model:
...
Repository: localaiLicense: apache-2.0
mimo-v2.6-distill-qwen-9b
MiMo-V2.6-Distill-Qwen-9B is Xiaomi MiMo's 9B Qwen3.5 fine-tune for coding, agent tasks, and visual coding.
This Q4_K_M GGUF build uses llama.cpp with the model's embedded chat template and includes the F16 vision projector.
Repository: localaiLicense: mit
mimo-v2.6-distill-qwen-9b-q8
MiMo-V2.6-Distill-Qwen-9B is Xiaomi MiMo's 9B Qwen3.5 fine-tune for coding, agent tasks, and visual coding.
This Q8_0 GGUF build uses llama.cpp with the model's embedded chat template and includes the F16 vision projector.
Repository: localaiLicense: mit
Attention: Trust Remote Code is required for this model
wemm-embedding-9b
WeMM-Embedding-9B is Tencent's largest Apache-2.0 multilingual embedding
model built on Qwen3.5. This entry serves the original bfloat16 safetensors
with LocalAI's Transformers backend and produces 4,096-dimensional
normalized embeddings for text retrieval, semantic search, and RAG.
The upstream model can also embed images and videos. LocalAI currently
exposes text input through its embeddings API for this backend.
Repository: localaiLicense: apache-2.0
ornith-1.0-9b-q4
Ornith-1.0-9B is an MIT-licensed Qwen3.5 model from Ornith AI for
agentic coding, reasoning, repository-level software tasks, and tool use.
It supports text and image input with a context window of 262K tokens.
This default entry uses the Q4_K_M GGUF and F16 vision projector. A
higher-quality Q8_0 model is available as a variant.
Repository: localaiLicense: mit
ornith-1.0-9b-q8
Ornith-1.0-9B in the higher-quality Q8_0 GGUF format, with the shared F16
vision projector for multimodal prompts.
Repository: localaiLicense: mit
qwen3.5-9b-kimi-k3-distilled-q4
A 9B Qwen3.5 fine-tune trained on Kimi K3 coding, debugging, and
tool-use traces. This Q4_K_M GGUF build uses llama.cpp and the embedded
chat template, with an F16 vision projector and a 32,768-token context.
Repository: localaiLicense: apache-2.0
qwen3.5-9b-kimi-k3-distilled-q8
A 9B Qwen3.5 fine-tune trained on Kimi K3 coding, debugging, and
tool-use traces. This Q8_0 GGUF build uses llama.cpp and the embedded
chat template, with an F16 vision projector and a 32,768-token context.
Repository: localaiLicense: apache-2.0
ornith-1.5-9b-q4
Ornith-1.5-9B is an MIT-licensed Qwen3.5 model from Ornith AI for
agentic coding, reasoning, repository-level software tasks, and tool use.
It supports text and image input with a context window of 262K tokens.
This default entry uses the Q4_K_M GGUF and BF16 vision projector.
Q5_K_M, Q6_K, and Q8_0 models are available as variants.
Repository: localaiLicense: mit
ornith-1.5-9b-q5
Ornith-1.5-9B in the Q5_K_M GGUF format, with the shared BF16
vision projector for multimodal prompts.
Repository: localaiLicense: mit
ornith-1.5-9b-q6
Ornith-1.5-9B in the Q6_K GGUF format, with the shared BF16
vision projector for multimodal prompts.
Repository: localaiLicense: mit
ornith-1.5-9b-q8
Ornith-1.5-9B in the higher-quality Q8_0 GGUF format, with the shared BF16
vision projector for multimodal prompts.
Repository: localaiLicense: mit
ornith-1.5-9b-uncensored-q4
Junafinity's Ornith-1.5-9B Uncensored is a 9B-parameter Qwen3.5
derivative with refusal-direction ablation for research and evaluation.
This Q4_K_M GGUF build includes the F16 vision projector and uses
the embedded chat template for text and image conversations.
Repository: localaiLicense: apache-2.0
ornith-1.5-9b-uncensored-q6
Junafinity's Ornith-1.5-9B Uncensored is a 9B-parameter Qwen3.5
derivative with refusal-direction ablation for research and evaluation.
This Q6_K GGUF build includes the F16 vision projector and uses
the embedded chat template for text and image conversations.
Repository: localaiLicense: apache-2.0
ornith-1.5-9b-uncensored-q8
Junafinity's Ornith-1.5-9B Uncensored is a 9B-parameter Qwen3.5
derivative with refusal-direction ablation for research and evaluation.
This Q8_0 GGUF build includes the F16 vision projector and uses
the embedded chat template for text and image conversations.
Repository: localaiLicense: apache-2.0
ornith-1.5-9b-obliterated-q4
Ornith-1.5-9B OBLITERATED is a refusal-removed derivative for alignment
research, red teaming, coding, reasoning, and agentic tasks. Its safety
guardrails are removed, and the publisher reports some capability loss
compared with the original model.
This default entry uses the Q4_K_M GGUF and BF16 vision projector. The
linked variant uses the higher-quality Q8_0 quantization.
Repository: localaiLicense: mit
Page 1