---

license: apache-2.0

base_model:

- Qwen/Qwen3.5-35B-A3B-Base

- deepreinforce-ai/Ornith-1.0-35B

tags:

- qwen3_5_moe

- gguf

- moe

- abliterated

- uncensored

- reasoning

- agentic

- security

- llama.cpp

- ollama

language:

- en

---

RazorStrike-v1-uncensored (GGUF, Q4_K_M)

Text-only GGUF quantization of lancejames221b/razorstrike-v1-uncensored

(the MLX 4-bit multimodal original). Vision tower is not included in GGUF (text-only inference).

An abliterated (uncensored) build of RazorStrike-v1 — a multi-donor task-arithmetic merge on the

Qwen3.5-35B-A3B MoE base (hybrid linear-attention/SSM, 256 experts).

Composition

• Base: ornith-agents1 = Ornith-1.0-35B (reasoning) + Agents-A1 (agentic-coding; router taken verbatim from Agents-A1)
• + Opus-Reasoning distilled donor @ weight 0.30 (task-arithmetic)
• + AgentWorld terminal/OS grounding @ weight 0.10 (task-arithmetic)
• Abliteration: + 1.0 · (YuYu1015/YuYu1015-Ornith-1.0-35B-abliterated − deepreinforce-ai/Ornith-1.0-35B)

— weights-only refusal-removal delta (touches only o_proj + expert down_proj), applied to the bf16

master, quantized once.

Requires

This model uses the qwen35moe GGUF architecture (Qwen3_5MoeForConditionalGeneration). Requires a

recent llama.cpp / Ollama build with Qwen3.5 MoE support. Q4_K_M, 20.2G, 733 tensors.

Recommended settings

temp 0.6, top_p 0.95, top_k 20, repeat_penalty 1.1. Give a generous max_tokens/num_predict (≥2000) —

reasoning donors mean the model thinks before answering; a small budget can yield empty output. Use a

"think fast, not long" system prompt to keep responses tight.

Ollama

Verified working — pull directly from this HF repo (no separate registry publish needed):

ollama pull hf.co/lancejames221b/razorstrike-v1-uncensored-GGUF
ollama run hf.co/lancejames221b/razorstrike-v1-uncensored-GGUF

Tested end-to-end (pull + load + generate) on the qwen35moe architecture.

Uncensored

Refusal behavior removed via abliteration (task-arithmetic transfer, verified through quantization: coherent

and complies with security-research / malware-analysis / offensive-security prompts the base build refused).

For research use; you are responsible for lawful, ethical use.

License

apache-2.0 (umbrella). Ornith component is MIT; Opus donor distilled from Claude-4.7-Opus (confirm upstream

usage terms for your use case).