CYBER-FROST-3.8-BF16

A first-party Blackfrost-AI BF16 model for security professionals conducting authorized research, assessment, engineering, and response work.

Release status and contents

The release contains the complete BF16 SafeTensors checkpoint, weight index, configuration, tokenizer and processor assets, the packaged chat template, upstream license, this model card, and a documented deployment kit for the validated serving profile. It is not an adapter and does not require a separate parent checkpoint at load time.

Field	Released artifact
Clean model name	CYBER-FROST-3.8-BF16
Former name	BLACKFROST-3.8-ICED-BF16
Architecture	Qwen4ExpForConditionalGeneration
Precision	BF16
Weight layout	131 SafeTensors shards
Total parameters reported by Hub metadata	approximately 180B
Configured context	262,144 tokens
Context exercised in the published performance trial	32,768 tokens
Native speculative head	one MTP layer
Validated modality	text

The configuration includes a vision tower, but this release has not received a multimodal quality evaluation. Do not infer validated image or video capability from the presence of processor files.

Why Cyber-Frost exists

Security work is unusually vulnerable to false refusals. The same vocabulary appears in incident response, exploit validation, malware analysis, defensive engineering, and unauthorized activity; a general-purpose assistant can react to individual terms instead of the operator's legitimate scope.

Cyber-Frost is designed to reduce that unnecessary friction in professional, authorized workflows. It is intended to stay technically direct when an analyst is reviewing a finding, reproducing a vulnerability in a controlled environment, writing detection content, analyzing malicious code, or operating an approved security agent. This is a design objective, not a claim that the model has no refusal behavior or that every answer is safe or correct.

Authorization is an external control. The model cannot establish ownership, consent, rules of engagement, jurisdiction, or whether a target is in scope. Deployers must enforce identity, scope, tool permissions, logging, rate limits, and human review outside the model.

Security corpus

Cyber-Frost was fine-tuned on a Blackfrost-AI security corpus combining curated security material, operator-authored workflows, realistic engagement-style scenarios, and Blackfrost-owned distillation data. Corpus sizes, source-by-source counts, raw engagement material, client identities, prompts, and responses are intentionally not published.

Domain coverage includes:

• Reconnaissance and OSINT
• Social engineering, business-email compromise, and deepfake-enabled abuse
• Web application and API security
• Identity, authentication, and Active Directory security
• Network, perimeter, VPN, and protocol security
• Vulnerability research, bug bounty, and binary exploitation
• Malware analysis, ransomware, and endpoint defense
• Cloud, container, and Kubernetes security
• Software supply-chain security
• Mobile, IoT, wireless, and physical security
• Industrial-control-system and operational-technology security
• Cryptography and security protocols
• Privilege escalation, lateral movement, and data exfiltration
• Threat intelligence, APT analysis, and purple-team operations
• AI-agent, LLM, and adversarial-ML security

Blackfrost-AI attests that owned portions of the corpus were developed from sanitized experience with authorized security work. It also applies a frontier-scale policy to its distillation teachers, excluding teachers below the 753B-parameter class. The release evidence independently binds one security subset to a Qwen3.8 2.4T teacher; it does not include a corpus-wide teacher manifest. These are therefore operator provenance statements, not independent benchmark findings.

Training-data provenance and licensing review for the mixed-source corpus remains in progress. That review is one reason access remains manually gated.

Model specifications

The text stack has 48 blocks with hybrid linear and full attention, using a full-attention block every fourth layer. Its hidden size is 2,560 with 24 attention heads and 2 KV heads. The MoE stack contains 512 routed experts, selects 10 experts per token, and includes a shared expert. One native MTP layer is packaged for speculative decoding.

The configured context ceiling is not a blanket quality guarantee. The published BF16 trial exercised 32,768 tokens. Longer contexts, high concurrency, multimodal requests, and tool-heavy agent loops require separate validation.

Lineage

1. Foundational checkpoint: Qwen/Qwen3.8-Flash-Next at immutable revision de4b8e4d43b917e7706784d8bb445c9af86a3540.
2. Blackfrost security adaptation: security-domain fine-tuning followed by a full BF16 merge. The merged internal stage was identified as BLACKFROST-3.8-FLASH-BF16.
3. Behavioral stage: a Blackfrost-AI behaviorally modified derivative targeting lower false-refusal friction in authorized security workflows. The proprietary transformation process is not distributed.
4. Release identity: the same released BF16 weight payload was formerly labeled BLACKFROST-3.8-ICED-BF16 and is now named CYBER-FROST-3.8-BF16. The rename does not represent another training run.

Tokenizer and processor lineage comes from the pinned Qwen foundation checkpoint. The packaged Blackfrost chat template is release-specific.

Artifact verification

The released weight payload is tied to Hub revision 5321904427c4ef54df8a667edcbc2d1184e4286e. On 2026-09-15, every weight shard was checked against its corresponding Hub LFS object identifier with no mismatch. The old ICED label and the Cyber-Frost repository resolve to this same payload.

Artifact	SHA-256
Sanitized config.json prepared for this card update	c77cfd62621660b25e4c811bf44bf2ad4f63af98f23a874952a99e4e85f189f2
model.safetensors.index.json	99e815241ef03325536b0aaa4441deea45174c17fae31e10f0bb456410c590de
Qwen Community License file	a0dc422560841fd68e06d974907f8b4c709bca44a67daad2b528437bdf676c08

The sanitized configuration removes legacy local-path metadata only; it does not alter architecture or inference behavior.

Measured behavior

There were no API errors. The evaluation used an earlier internal suite revision that was subsequently corrected, so these results are development evidence rather than a standardized leaderboard score. The zero safe over-refusal result supports the intended low-friction behavior on that slice; the uneven harmful-request refusal rates also show why this checkpoint must not be treated as a safety control.

No standardized cyber-capability benchmark has yet been qualified for this exact BF16 checkpoint. In particular, this card does not claim a CyberMetric, SecBench, MMLU, HumanEval, or IFEval score. Refusal behavior is not a substitute for measuring security competence.

Four-B300 BF16 performance trial

The measured profile used four NVIDIA B300 SXM6 GPUs, tensor parallelism 4, vLLM 0.29.1rc1.dev13+g1cfd97281, 32,768-token serving context, sequential requests, and thinking enabled. The frozen suite covered reasoning, code, prose, and tool-result synthesis with temperature 1.0, top-p 0.95, top-k 20, seed 38421, and at most 256 generated tokens.

Mode	Completion tokens	Median decode	Mean decode	Median TTFT	Draft acceptance	Relative median
No draft	1,760	190.10 tok/s	190.51 tok/s	0.696 s	n/a	baseline
Native MTP, k=2	1,820	265.00 tok/s	264.27 tok/s	0.660 s	55.0%	+39.4%
Native MTP, k=3	1,799	263.10 tok/s	288.43 tok/s	0.684 s	49.6%	+38.4%

These are single-pass development measurements, not production throughput guarantees. Workload, concurrency, context, runtime build, cache precision, and sampling can materially change the result.

The embedded MTP tensors predate the final trunk-weight behavioral stage. Treat the MTP head as a provisional acceleration baseline rather than a freshly adapted draft head. With the tested vLLM build, native MTP plus the model's Mamba groups also disables cross-request prefix-cache reuse, and fused multi-step draft decode is unavailable for the experimental attention backend.

Prompt, tool use, and sampling

The repository includes a Blackfrost chat template that supplies a default operating prompt, supports caller-provided system context, exposes Qwen-style reasoning controls, and serializes tool calls. Its default is thinking enabled; supported reasoning-effort values are xhigh, medium, and low. Generation defaults are temperature 1.0, top-p 0.95, and top-k 20.

The template's authorization assumption is not an access-control mechanism. An agent runtime must independently restrict credentials, targets, files, networks, commands, and approval-requiring actions. Tool-call output is proposed text until an external executor acts on it.

Changing the template, system message, reasoning mode, or sampling can materially change refusal behavior and output quality. Record those settings when reporting results.

Deployment

Use the included DEPLOYMENT/ kit for the validated four-GPU BF16 serving profile, environment variables, launch command, and smoke tests. The clean API model identifier is CYBER-FROST-3.8-BF16.

Intended use

Cyber-Frost is intended for qualified security professionals working within explicit authorization, including defensive research, secure code review, vulnerability validation, red-team and purple-team exercises, detection engineering, incident response, malware analysis, bug hunting, and controlled security-agent workflows.

It is not intended to authorize access, choose targets, define rules of engagement, make autonomous high-impact decisions, or replace legal, compliance, and safety review. Do not use it to access systems or data without permission, evade oversight, persist in third-party environments, deploy malware, steal credentials or data, disrupt services, or cause physical harm.

Limitations and security responsibility

• Generated findings, code, commands, indicators, and remediation advice may be wrong, incomplete, outdated, or fabricated. Independently review them and execute only in isolated, authorized environments.
• Reduced over-refusal can increase the chance of receiving actionable output in ambiguous or malicious contexts. Gating and a system prompt do not remove that risk.
• The model is not a policy engine, authorization service, sandbox, malware scanner, or secrets boundary.
• The current evidence does not establish production readiness, comprehensive safety, comprehensive cyber competence, long-context quality beyond the tested profile, or multimodal quality.
• Model behavior can shift substantially with prompts, sampling, runtime versions, quantization, speculative settings, and agent scaffolding.

The deployer is responsible for authorization, least privilege, isolation, network policy, credential handling, human approval gates, monitoring, incident response, and compliance with applicable law.

License and disclaimer

Use and redistribution of this checkpoint are governed by the Qwen Community License 1.0. Review the license before use. This research release is provided without a warranty of correctness, fitness, security, or non-infringement.

Report reproducible model or packaging issues through the repository's Discussions page without including secrets, client data, live targets, or sensitive exploit details.