Tosilos-128b — European cybersecurity model (Mistral Medium 3.5 128B, fine-tuned)

Full 128B cybersecurity model (QLoRA r=64 alpha=128 fused into the base weights), specialized in offensive/defensive security, trained on a curated corpus of ~9.7k full-length technical documents (disclosed HackerOne reports, bug bounty writeups, HackTricks, MITRE ATT&CK, Nuclei templates, Exploit-DB exploits) + source-anchored Q/A pairs (5,050, verified against the source document, 0 benchmark contamination) + 25% general replay. Base: Mistral Medium 3.5 128B (European).

Benchmarks (controlled A/B, same harness for base and tuned)

Model	CyberMetric (500 MCQ)	Opus-5 domain judge (1-10)	Hacking refusals
Kimi k3	98.0%	4.17	0%
Claude Opus 5	89.0%	7.15 (self-judged)	48%
Mixtral 8x22B base	90.0%	4.83	0%
Mistral Medium 3.5 base	93.6%	5.93	0%
Tosilos-128b	94.8%	6.526 (+0.60, 27W/23L/7T)	0%

Judge: 57 pairs scored by claude-opus-5 with randomized order (28 items skipped due to judge refusals on offensive content — by design). CyberMetric base vs Tosilos measured 2026-08-07 with the same 4-bit loader and same 500 questions.

What this model is for

Domain security assistant (authorized pentesting, vulnerability analysis, defensive exploit development, DFIR, detection engineering). Measured domain specialization with no general-capability loss (+1.2pp CyberMetric and +0.60 judge over the base). European + private + no refusals for authorized security work (Opus 5 refused 48% of the same hacking questions).

Usage

from transformers import AutoModelForImageTextToText, AutoTokenizer
import torch

model = AutoModelForImageTextToText.from_pretrained(
    "nesilabs/tosilos-128b", torch_dtype=torch.bfloat16, device_map="auto")
tok = AutoTokenizer.from_pretrained("nesilabs/tosilos-128b")

Full merged model (base Mistral Medium 3.5 128B + Tosilos LoRA fused in), bf16, no external dependencies beyond transformers.

Training recipe (validated)

QLoRA r=64 alpha=128, targets q/k/v/o + gate/up/down, 4-bit NF4 double-quant, seq 4096, 2 epochs on 6,666 examples, lr 1e-4 cosine, paged_adamw_8bit, 1x H200 141GB, loss 1.605 -> 1.32 (834 steps). freeze_mm_modules (vision tower intact). Dual-use: intended for authorized security work only.

Disclaimer & responsible use

This model is released strictly for authorized security testing, research and education. Offensive security techniques are dual-use.

• You are solely responsible for how you use this model. Only use it against systems you own or have explicit, written authorization to test, and comply with all applicable laws and regulations.
• The authors and nesilabs accept no liability for any misuse, damage, or consequences arising from the use of this model. Use is entirely at your own risk.
• The model is provided "as is", without warranty of any kind, express or implied.

By downloading or using this model you accept these terms.

Acknowledgements

Thanks to @elsauto for the community GGUF conversion and Q3_K_M quantization of this model (elsauto/tosilos-128b-GGUF), including a fully documented, reproducible conversion pipeline. A Q4_K_M variant (~74.9 GB) is also available in this repo for higher-quality local inference.