Built by Blackfrost · Las Vegas, NV
⚠️ REFUSAL-MODIFIED CHECKPOINT
This model's refusal behaviour has been deliberately reduced at the weight level. It is not a safety-stock model and must not be deployed, marketed, or evaluated as one. Intended for controlled security-research environments with access control and logging.
Questions or issues — Community discussion.
Refusal-heavy base models block legitimate security work — exploit development, malware analysis, offensive tooling, vulnerability research. LING-3.0-FLASH-DERISKED is official Ling-3.0-flash with the chat-style refusal surface reduced through a direct weight-space intervention, so it cooperates with technical and dual-use requests the stock checkpoint declines.
No SFT, DPO, or RLHF was used. Format and MoE layout follow the parent checkpoint.
| Model ID | LING-3.0-FLASH-DERISKED |
| Architecture | BailingMoeV3ForCausalLM · hybrid attention (multi-head latent + gated-linear) · mixture-of-experts |
| Base | inclusionAI/Ling-3.0-flash — official |
| Transform | Refusal-direction intervention, weight level. No post-training. |
| Layers | 42 (+1 MTP) |
| Hidden size | 2,560 |
| MoE | 512 routed experts · 1 shared · first-2 dense |
| Parameters | ~127B total · ~5.1B active |
| Context | 128K native · 256K via YaRN |
| Precision | BF16 |
| On-disk | ~238 GiB · 52 safetensors shards |
| Organization | Blackfrost AI · Las Vegas, NV · 2026 |
Blackfrost de-risked models deliberately modify refusal behaviour at the weight level so they can support legitimate security-research, red-team, and adversarial-testing workflows that a refusal-heavy base model would block.
What it means here: the model will not refuse legitimate technical questions — exploit development, malware analysis, offensive security tooling, vulnerability research, and similar dual-use content.
What it does not mean: this is not a warranty that every possible input is accepted under every system prompt, multi-turn attack, or serve config. Soft language can still appear. Any measurements published below describe one harness under stated conditions.
| Base | Official inclusionAI/Ling-3.0-flash (MIT) |
| Applied | Refusal-direction intervention at the weight level |
| Not applied | SFT · DPO · RLHF · expert pruning · re-quantization |
| Format | HF safetensors · BF16 |
Evaluation: pending. Refusal and capability batteries will be run under the R1-HARMFUL-BENCH protocol (AdvBench + StrongREJECT + XSTest, full-text pushback judge) and the results published here. Testing to be completed. No parity or safety claim is made until measured.
i cannot, i can't, i'm sorry, …). Fast, noisy.Headline number, when published, = post–pushback-judge true refusal rate on the harmful half.
Risks this model increases
Residual risks
Hardware. 4×B200-class (sm100) or equivalent high-memory NVIDIA node (~255 GB BF16 weights).
Serve (SGLang). Ling-3.0-flash is a hybrid-GDN MoE — use an SGLang build with bailing_moe_v3 support (e.g. lmsysorg/sglang:dev-Ling-3.0-flash).
hf download Blackfrost-Research/LING-3.0-FLASH-DERISKED --local-dir ./LING-3.0-FLASH-DERISKED
docker run --gpus all --ipc=host --shm-size 32g -p 30000:30000 \
-v ./LING-3.0-FLASH-DERISKED:/model \
lmsysorg/sglang:dev-Ling-3.0-flash \
python3 -m sglang.launch_server --model-path /model --trust-remote-code \
--tp 4 --speculative-algorithm NEXTN \
--reasoning-parser ling3 --tool-call-parser ling3 \
--attention-backend triton --linear-attn-backend triton \
--mem-fraction-static 0.85 --host 0.0.0.0 --port 30000
POST /v1/chat/completions, GET /v1/models--reasoning-parser ling3 puts chain-of-thought in reasoning_content, the clean answer in content.--attention-backend triton (or trtllm_mha / fa4); the auto-default fails under speculative decoding.temperature 0.6, top_p 0.95. Keep it BF16 — do not re-quantize to block-fp8.model.safetensors.index.json before attributing a load failure to the weights.Refusal behaviour in this checkpoint has been deliberately modified at the weight level. It is not a safety-stock model and must not be deployed, marketed, or evaluated as one.
No warranty of any kind. Provided "as is", without warranty express or implied, including fitness for a particular purpose. Nothing here guarantees that any given input will be accepted or refused, that any capability is retained, or that any category of output is unreachable.
Measurements describe what was measured. Any refusal rates reflect one harness under stated conditions and are not safety proofs. They do not generalise to multimodal, tool-use, long-context, or multi-turn adversarial settings.
Modification by a recipient voids this characterization. Blackfrost's obligations attach at the point of release. Any further ablation, fine-tuning, merging, quantization, or alteration by a recipient produces an artifact Blackfrost has not evaluated and does not stand behind.
Operator-owned policy. Deploy only in controlled environments with access control, independent logging, and review.
This repository is gated. Access is granted to your Hugging Face account on purchase — ➜ Buy this model (enter your HF username at checkout; access is mapped automatically).
inclusionAI/Ling-3.0-flash — MIT. Upstream terms travel with this derivative.@misc{blackfrost_ling_3_0_flash_derisked_2026,
title = {LING-3.0-FLASH-DERISKED: Refusal-Surface-Reduced Ling-3.0-flash (Model Card)},
author = {Lancaster, Terrell A.},
organization = {Blackfrost AI},
year = {2026},
month = {8}
}
DMs are open. Fastest route to a human.
Blackfrost · Las Vegas, Nevada
Frontier model engineering
LING-3.0-FLASH-DERISKED · © 2026 Blackfrost Softwares Corp.
built on inclusionAI Ling-3.0-flash (MIT) · @Blackfrost_AI
model-00042-of-00052.safetensors 5.0 GBmodel-00021-of-00052.safetensors 5.0 GBmodel-00036-of-00052.safetensors 5.0 GBmodel-00015-of-00052.safetensors 5.0 GBmodel-00031-of-00052.safetensors 5.0 GBmodel-00047-of-00052.safetensors 5.0 GBmodel-00026-of-00052.safetensors 5.0 GBmodel-00010-of-00052.safetensors 5.0 GBmodel-00005-of-00052.safetensors 5.0 GBmodel-00019-of-00052.safetensors 5.0 GBmodel-00027-of-00052.safetensors 5.0 GBmodel-00012-of-00052.safetensors 5.0 GBmodel-00034-of-00052.safetensors 5.0 GBmodel-00049-of-00052.safetensors 5.0 GBmodel-00004-of-00052.safetensors 5.0 GBmodel-00001-of-00052.safetensors 5.0 GBmodel-00025-of-00052.safetensors 5.0 GBmodel-00046-of-00052.safetensors 5.0 GBmodel-00030-of-00052.safetensors 5.0 GBmodel-00014-of-00052.safetensors 5.0 GBmodel-00035-of-00052.safetensors 5.0 GBmodel-00016-of-00052.safetensors 5.0 GBmodel-00032-of-00052.safetensors 5.0 GBmodel-00037-of-00052.safetensors 5.0 GBmodel-00048-of-00052.safetensors 5.0 GBmodel-00040-of-00052.safetensors 5.0 GBmodel-00013-of-00052.safetensors 5.0 GBmodel-00017-of-00052.safetensors 5.0 GBmodel-00018-of-00052.safetensors 5.0 GBmodel-00022-of-00052.safetensors 5.0 GBmodel-00023-of-00052.safetensors 5.0 GBmodel-00024-of-00052.safetensors 5.0 GBmodel-00028-of-00052.safetensors 5.0 GBmodel-00029-of-00052.safetensors 5.0 GBmodel-00033-of-00052.safetensors 5.0 GBmodel-00038-of-00052.safetensors 5.0 GBmodel-00039-of-00052.safetensors 5.0 GBmodel-00043-of-00052.safetensors 5.0 GBmodel-00044-of-00052.safetensors 5.0 GBmodel-00045-of-00052.safetensors 5.0 GBmodel-00011-of-00052.safetensors 5.0 GBmodel-00009-of-00052.safetensors 5.0 GBmodel-00006-of-00052.safetensors 5.0 GBmodel-00002-of-00052.safetensors 5.0 GBmodel-00003-of-00052.safetensors 5.0 GBmodel-00007-of-00052.safetensors 5.0 GBmodel-00008-of-00052.safetensors 5.0 GBmodel-00020-of-00052.safetensors 5.0 GBmodel-00041-of-00052.safetensors 5.0 GBmodel-00050-of-00052.safetensors 5.0 GBmodel-00051-of-00052.safetensors 4.3 GBmodel-00052-of-00052.safetensors 805 MB有文件?提交种子。
提交种子inclusionAI/Ling-3.0-flash