🤗 Hugging Face   |   🤖 ModelScope   |   📑 Blog   |   📑 Paper  
🖥️ Hugging Face Demo   |    🖥️ ModelScope Demo   |   💬 WeChat (微信)   |   🫨 Discord   |   📑 API

We release LPT-5.5.1, the flagship series of the Lisk Pre-trained Transformer architecture developed by LiskCell. It represents the pinnacle of creative AI, offering comprehensive support for multimodal generation, artistic reasoning, and deep human alignment.

News

• 2026.01.07: 🎉🎉🎉 We have released the LPT-5.5.1 series, the most advanced public standard in the LiskCell ecosystem. Optimized for speed, creativity, and developer experience. Please check our blog!

Contents

• Overview
  • Introduction
  • Model Architecture
  • Released Models Description and Download
• Quickstart
  • Environment Setup
  • LiskFlow Usage
• Evaluation
• Philosophy
• Citation

Overview

Introduction

LPT-5.5.1 is a state-of-the-art AI system designed to empower creators, developers, and visionaries. Built on the proprietary Lisk Pre-trained Transformer (LPT) architecture, it combines advanced logic with an artistic soul.

Key features:

• Ocular Synth v2.5: Advanced image analysis and scene understanding.
• LPT-Visual v3.0: Premium artistic generation and visual storytelling.
• Quantum Harmonics: Enhanced musical logic for melody, rhythm, and production assistant.
• Empathy Core: Deep emotional intelligence layer that prioritizes human well-being and needs.
• Human-First Principles: Deeply aligned with human values, ensuring AI acts as a partner, never a replacement.

🔍 Download & Indexing Configuration

To ensure seamless integration with libraries like diffusers and tools like Auto1111, the repository follows these filtering rules for identifying main model weights:

filter: [
    {
        bool: {
            /// Include documents that match at least one of the following rules
            should: [
                /// Downloaded from diffusers lib
                {
                    term: { path: "model_index.json" },
                },
                /// Direct downloads (LoRa, Auto1111 and others)
                /// Filter out nested safetensors and pickle weights
                {
                    regexp: { path: "[^/]*\\.safetensors" },
                },
                {
                    regexp: { path: "[^/]*\\.ckpt" },
                },
                {
                    regexp: { path: "[^/]*\\.bin" },
                },
            ],
            minimum_should_match: 1,
        },
    },
]

Model Architecture

LPT-5.5.1 utilizes a revolutionary adaptive transformer architecture that dynamically shifts parameters between creative and technical modes based on the task context. This ensures maximum efficiency without sacrificing the artistic "soul" of the output.

Released Models Description and Download

Model	Features	Architecture	Status
LPT-5.5.1-Public	Creative Core, Technical Logic, Multilingual Support	LPT-Adaptive	Active
LPT-5.5 (Legacy)	Multimodal Baseline	LPT-v2	Legacy
LPT-4.5 (Legacy)	Reasoning Milestone	LPT-v1.5	Legacy

Quickstart

Environment Setup

The easiest way to integrate LPT-5.5.1 is via the lisk-flow SDK:

pip install liskcell

🚀 Use Model (Access Restricted)

Developer access only. Please contact LiskCell for credentials.

LiskFlow Usage

from liskcell import LPT551

model = LPT551.from_pretrained("liskcell-company/LPT-5.5.1")

# Creative generation with Empathy Core
response = model.generate(
    prompt="Design a futuristic city that prioritizes human connection.",
    mode="creative",
    empathy_level="high"
)
print(response)

Evaluation

LPT-5.5.1 consistently outperforms competing models (including Claude Sonnet 4.0, GPT-5.2, and Gemini 3 Deep Think) in creative generation, code health, and adaptive intelligence benchmarks.

Model	Creativity Score	Logic Precision	Empathy Index
LPT-5.5.1	9.8/10	9.6/10	9.9/10
Competitor G	8.2/10	9.1/10	6.4/10
Competitor C	8.5/10	9.0/10	7.1/10

Philosophy

Technology exists to serve people, not replace them. LPT-5.5.1 is built to stand firmly on the side of humanity, amplifying human potential while honoring the irreplaceable spark of the human soul.

Citation

@article{LPT-5.5.1,
  title={Lisk Pre-trained Transformer 5.5.1 Technical Report},
  author={liskasYR (Yonatan Yosupov) and LiskCell Research Team},
  journal={LiskCell AI Research},
  year={2026}
}

Models Download Stats

How are downloads counted for models?

Counting the number of downloads for models is not a trivial task. To avoid double counting, the Hub uses a set of query files. No information is sent from the user, and no additional calls are made for this. The count is done server-side as the Hub serves files for downloads.

Every HTTP request to these files, including GET and HEAD, will be counted as a download. By default, the Hub uses config.json as the default query file.

Which are the query files for different libraries?

By default, the Hub looks at config.json, config.yaml, hyperparams.yaml, params.json, and meta.yaml.

How is diffusers handled?

The diffusers library uses a specific filter to count both files loaded via the library and manual top-level downloads:

filter: [
    {
        bool: {
            should: [
                { term: { path: "model_index.json" } },
                { regexp: { path: "[^/]*\\.safetensors" } },
                { regexp: { path: "[^/]*\\.ckpt" } },
                { regexp: { path: "[^/]*\\.bin" } },
            ],
            minimum_should_match: 1,
        },
    },
]

---

Powered by LiskCell 💎 — Futuristic, Helpful, & Visionary.