Get the latest setups
Models/GLM 5.3 CYBERSECURITY FP8
Text / dealignai

GLM 5.3 CYBERSECURITY FP8

GLM 5.3 CYBERSECURITY FP8 is a text-generation model published at dealignai/GLM-5.3-CYBERSECURITY-FP8 under the MIT license. It is described as a cybersecurity-domain, refusal-reduced ('abliterated'/'crack') FP8 variant based on zai-org/GLM-5.3 and JANGQ-AI/GLM-5.3-FP8. The model card states it targets offensive-security, red-team, exploit-dev, reverse-engineering, evasion, phishing, credential-attack, and malware-analysis content, with refusal reduction noted as domain-specific rather than general-purpose. It is a 753B total parameter, text-only glm_moe_dsa MoE model with 78 layers, using routed FP8 experts with edited bf16 residual writers, and is intended for native FP8 tensor-core speed on Hopper GPUs. Supported languages listed are en, zh, ru, sr, hi, fr, es, ar, ko, and ja.

EstimatedRepository opened Aug 30, 2026Source checked 9/23/2026Version: 5915c1b8
LOCALRENTED GPUOPEN WEIGHTS
Freshness 76 out of 100. Repository age: 70 days old, Aug 30, 2026. Measures age, not quality.
Popularity 70 out of 100. 127,638 downloads in 30 days; 881 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
What do these scores mean?

Two ways to compare models. Neither score measures answer quality, benchmark performance, or whether a model fits your machine.

Freshness · release age

A new release starts at 100. The score halves every 180 days: 100 × 2^(−age in days / 180). When the release date is unavailable, a repository date is a fallback—not proof of when the model was released. Missing dates stay unscored.

Check the date source
Popularity · adoption

Downloads over the past 30 days carry 80% of the score; lifetime likes carry 20%. Both use a logarithmic scale, capped at 10 million downloads and 100,000 likes. If likes are unavailable, downloads carry all the weight. Missing download statistics stay unscored.

round(80 × min(1, log10(1 + downloads) / 7) + 20 × min(1, log10(1 + likes) / 5))

Check the statistics source
Scores as of Nov 8, 2026. Statistics checked Nov 7, 2026. Statistics are marked for refresh after 72 hours.Full scoring method

At a glance

Parameters
753.33B
Architecture
glm_moe_dsa
Context length
1,048,576
License
mit
Disk space
703.7 GB
Software
vLLM
View the model source
MY HARDWARE

Will it run on your machine?

Save your machine to see a personalized rating and its reasoning.

Add my hardware

Best for

  • Offensive-security, red-team, and penetration-testing content
  • Exploit development and reverse-engineering tasks
  • Evasion, phishing, and credential-attack-related content
  • Malware analysis and cybersecurity-focused technical assistance
  • Multilingual cybersecurity text generation (en, zh, ru, sr, hi, fr, es, ar, ko, ja)

Tradeoffs

  • Refusal reduction is specifically tuned for cybersecurity and is not a general-purpose uncensor
  • Non-cyber categories such as weapons, chemistry, biology, harassment, and misinformation may still receive soft educational-style refusals
  • Copyright-verbatim reproduction still soft-refuses in this variant
  • reasoning_effort reportedly only honors 'low' and 'high', with other values falling through to 'max'; reasoning cannot be disabled on this checkpoint
  • At high/max reasoning, the model can exhaust max_tokens inside reasoning and return zero answer tokens; practical TP8 H200 context ceiling is reported around 131K with MTP and about 160K without

Ways to run it

vLLM · Linux

Repository-specific command in publisher documentation; confirm dependencies and hardware in the source.

vllm serve dealignai/GLM-5.3-CYBERSECURITY-FP8 --tensor-parallel-size 8 --gpu-memory-utilization 0.90 --enforce-eager --disable-custom-all-reduce --enable-prefix-caching --max-num-seqs 24 --max-model-len 131072 --reasoning-parser glm45 --tool-call-parser glm47 --enable-auto-tool-choice
Official instructions

Explore its uses

SAFETENSORSTEXT-GENERATION

Keep exploring

Text
Prism ML

Bonsai 2 · 27B

A compressed 27B-class reasoning model with publisher-provided GGUF packs and a dedicated llama.cpp fork for CUDA, Metal, and CPU.

Freshness 82 out of 100. Release age: 52 days old, Sep 17, 2026. Measures age, not quality.
Popularity 85 out of 100. 1,710,588 downloads in 30 days; 2,952 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
MODEL SIZE27.36B
LOCALRENTED GPU
Released Sep 17, 2026
Text
DeepSeek

DeepSeek V4.1 Flash

An open-weight model for reasoning, coding, agents, and image understanding. DeepSeek V4.1 Flash combines a 552B backbone with sparse activation and up to one million tokens of context. Choose the original checkpoint or a community uncensored variant below.

Freshness 80 out of 100. Repository age: 59 days old, Sep 10, 2026. Measures age, not quality.
Popularity 87 out of 100. 1,911,780 downloads in 30 days; 4,887 lifetime likes. Checked Nov 7, 2026. Measures adoption, not quality.
MODEL SIZE552B backbone · 8B active prefill / 16B active decode
LOCALRENTED GPU
Repository opened Sep 10, 2026
Text
Qwen

Qwen3.8

Explore 4 versions of Qwen3.8. Compare model sizes, packages, hardware requirements, and available setups.

4 versionsCompare sizes & formats
LOCALRENTED GPU
Compare versions →