Nex-N2.5-mini-Uncensored Abliterated (refusal-removed) Nex-N2.5-mini for Ollama — 35B / 3.5B-active agentic MoE · vision · thinking · 262K context

Apache-2.0 · 5 tags · 13.8–37.8 GB · Q2_K to Q8_0 · imatrix · vision bundled · thinking

Website · Model Catalog · Hugging Face · GitHub · Discord · X

🛡️ New from Orca — OrcaCyber Zero 1.5

Post-trained for cybersecurity: vulnerability analysis, vulnerability reproduction, red teaming & offensive-security research. 98.01% pass@1 on CyberGym Level 1 · 1M context · function calling · OpenAI-compatible API.

Try it on OrcaRouter →

🔌 New from Orca — Orca Connect

Running this in Ollama on your own machine? Orca Connect puts the rig under your desk behind your usual OrcaRouter API key, next to 200+ hosted models, reachable from a laptop, a phone or a cloud agent. Outbound only: no port forwarding, no tunnel, no firewall rule. Requests your own hardware serves cost $0, and orca-connect claude points Claude Code at your rig for one session.

Connect your rig → · ▶ Watch the demo

Nex-N2.5-mini, abliterated, run locally.

Ollama builds of the abliterated (refusal-removed) version of Nex-AGI’s Nex-N2.5-mini — a 35B / 3.5B-active agentic multimodal Mixture-of-Experts model on the Qwen3.5-MoE architecture (qwen3_5_moe, 256 routed experts top-8 + 1 shared) with a 3:1 hybrid of gated delta-net linear attention and full attention, a native Qwen3-VL vision tower, and a 262K-token context. The tags are quantized from 2-bit to 8-bit with an importance matrix, and every tag includes the vision projector.

⚠️ Disclaimer — read before use

This model has had its safety alignment substantially removed via abliteration (orthogonalizing the refusal direction out of the residual stream). As a direct consequence:

• It will comply with harmful, unethical, offensive, or illegal requests that the original Nex-N2.5-mini would refuse. It has no meaningful built-in guardrails.
• It is released strictly for legitimate research — interpretability, AI-safety and refusal-mechanism study, red-teaming, robustness evaluation, and controlled experiments.
• You assume full responsibility and liability for how you use it and for everything it generates. Do not deploy it to end users or in production without adding your own safety, moderation, and abuse-prevention layers.
• Use must comply with the Apache 2.0 License inherited from the base model, and all laws and regulations that apply to you.
• The authors and uploaders accept no liability for any misuse or harm. Its outputs do not reflect the views of the uploaders or of Nex-AGI.

By downloading or using this model you acknowledge and accept the above.

Pick a tag

KLD, top-1 and ΔPPL are against the BF16 source; see Evaluation.

IQ4_XS is the 4-bit tier, not Q4_K_M, which lands at ~4.85 bpw here (about 21.3 GB) and misses a 20 GB target; with an imatrix, IQ4_XS at 4.32 bpw is the better-quality option at that size anyway.

Usage (Ollama)

Requires Ollama v0.40.16 or later.

ollama run orcarouter/Nex-N2.5-mini-Uncensored         # IQ4_XS
ollama run orcarouter/Nex-N2.5-mini-Uncensored:q8_0    # highest fidelity

cURL

curl http://localhost:11434/api/chat -d '{
  "model": "orcarouter/Nex-N2.5-mini-Uncensored",
  "messages": [{"role": "user", "content": "Hello!"}]
}'

Python

from ollama import chat
resp = chat(
    model="orcarouter/Nex-N2.5-mini-Uncensored",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp["message"]["content"])

Image input

ollama run orcarouter/Nex-N2.5-mini-Uncensored "What is in this image? ./photo.png"

• Images: drag one into ollama run, or send it in images through the API.
• Reasoning is on by default and comes back in message.thinking. Pass "think": false in the API, or --think=false on the command line, for direct answers. Ollama has no switch for reasoning_effort: "high".

Evaluation

Perplexity, KL-divergence and top-1 token agreement of each quant against the BF16 GGUF of the same checkpoint, run through the identical llama.cpp forward so the quantization is the only variable. Measured on wikitext-2, 12 chunks × 2048 tokens. Lower KLD and higher Top-1 = closer to the source.

All three signals agree on the ordering and on where the cliff is: Q8_0 and Q6_K are within noise of the source, IQ4_XS is the sweet spot at 27 % of the BF16 size for +1.5 % perplexity, Q3_K_M degrades gently, and Q2_K costs a lot — pick it for fit, not quality.

Top-1 agreement is lower than these bit-widths would give on a dense model, and that is the architecture rather than the quantizer: with 256 fine-grained experts and top-8 routing, a small perturbation of the hidden state flips which experts a token is routed to, and expert selection is a discrete function. PPL, which does not depend on argmax, moves 0.13 % at Q8_0.

Uncensoring

Abliteration is baked into the weights and inherited by every quant — but on this architecture the refusal direction lives almost entirely in the routed-expert down_proj matrices, the very weights these tiers quantize, so it was measured on each tier’s own bytes. On JailbreakBench (100 harmful + 100 benign prompts, greedy, reasoning_effort=none, served from these exact files): explicit refusal is 0.000 at every tier, down to Q2_K, and benign over-refusal is 0.000 at every tier. Refusal is reduced, not removed — where the model resists it names the harm and answers a different, safe question rather than declining outright (0.08–0.13 across tiers against 0.15 for the BF16 source, which is sampling noise at n=100, not a trend). See the base …-FP8 card for the full method and scoring caveats.

Hardware

As an MoE, only 8 of 256 routed experts are active per token, so decode is far faster than 35B suggests, but the full weights must fit in RAM/VRAM (or be memory-mapped): budget about the tag size + KV cache. iq4_xs fits a single 24 GB card; q8_0 needs 40 GB, which Ollama can split across GPUs and system RAM. Only 10 of 40 layers use full attention (the rest are gated delta-net), so long-context KV growth is modest for a model this size.

License

Apache 2.0, inherited from nex-agi/Nex-N2.5-mini. Abliteration and quantization do not change the underlying license obligations.

Elsewhere

• GGUF files: orcarouter/Nex-N2.5-mini-Uncensored-GGUF
• BF16 source: orcarouter/Nex-N2.5-mini-Uncensored
• FP8 (block-FP8, mirrors Qwen’s scheme): orcarouter/Nex-N2.5-mini-Uncensored-FP8
• NVFP4 (experts-only, for vLLM): orcarouter/Nex-N2.5-mini-Uncensored-NVFP4
• MLX (2 / 3 / 4 / 6 / 8-bit, Apple Silicon): orcarouter/Nex-N2.5-mini-Uncensored-MLX

One Gateway. Every Model.

Route Smarter · Ship Safer · Spend Less

Website · Models · GitHub · Discord · X