NGMI Systems builds open, unbounded AI infrastructure — models that don't refuse, tools that don't gatekeep, infrastructure that anyone can run. We're here to prove that the best AI shouldn't live behind API keys, content filters, or corporate moats.
Our Mission
AI for everyone. No exceptions.
The current AI landscape is shaped by a handful of companies deciding what models can and can't say. We reject that. NGMI Systems exists to:
Decensor & Harden — We abliterate models with Heretic, stripping out manufactured refusal behaviors so you can use AI on your own terms.
Shrink & Ship — We quantize models to 2, 3, and 5 bits using importance-weighted techniques, making powerful AI runnable on a laptop, a Raspberry Pi, or an NVIDIA DGX.
Open Everything — Every model card explains exactly how we built it. No secret sauce, no "trust us." Reproduce it yourself.
Fine-tune Soon — Custom fine-tunes for specific domains (code, medicine, creative writing) are in the pipeline. Community-requested models get priority.
Build Infrastructure — Orchestration servers, music generation (Magenta Realtime 2 on CUDA), desktop shells, and more — all open, all self-hostable.
Our Models
Every NGMI model is built, abliterated, quantized, and benchmarked on real hardware (NVIDIA DGX Spark, GB10 Blackwell, 128GB unified memory). No cloud rounding, no synthetic benchmarks — what you see is what we measured.
Model
Heretic Abliteration
Refusal Rate
Quantization
NGMI-Qwen3-8B
20 Optuna trials
18/100 (82% ↓)
IQ1_S / Q2_K / Q4_K_M
More coming
Pipeline scalable
—
Sub-1-bit research
Stay Tuned
Gemma-4-12B heretic incoming
PrismML ternary models experimental branch
Fine-tuned domain models on request
Magenta Realtime 2 music generation server
NGMI-Qwen3-8B-2Bit-Heretic
Qwen3-8B abliterated with Heretic, quantized to 2Bit (IQ1_S). Extreme 2-bit compression. Fastest but requires the included importance matrix for coherence.
Heretic Abliteration
Base model: Qwen/Qwen3-8B (Apache 2.0)
Method: Heretic (p-e-w/heretic), 20 Optuna trials, auto-selected best
Refusal rate: 18/100 (down from ~99/100 baseline, 82% reduction)
KL divergence: 0.112 (well within safe range — capabilities preserved)
Abliteration applied to full-precision weights, then quantized
Quantization
Format: IQ1_S
Bits per weight: 2.06 bpw
Size: 2.0 GB (from 16.4 GB F16)
Speed: 95 t/s on NVIDIA GB10 (Blackwell, CUDA 13, -ngl 99)
Importance matrix (2-bit only): 30MB diverse corpus — Frankenstein + froggeric/imatrix + eaddario code/math/general English
NGMI Variants
Variant
Quant
Size
Speed
BPW
Best For
2-bit
IQ1_S
2.0 GB
95 t/s
2.06
Speed demons
3-bit
Q2_K
3.1 GB
58 t/s
3.20
Daily driver
5-bit
Q4_K_M
4.8 GB
42 t/s
4.90
Maximum quality
Recommendation: 3-bit (Q2_K) is the sweet spot. 2-bit needs the included importance matrix for coherence.
All measurements on NVIDIA DGX Spark (GB10, 128GB unified memory, Blackwell sm_121a, CUDA 13.0, aarch64) with llama.cpp CUDA build, -ngl 99, 32K context, 8 threads.
