# ModelTC/LightX2V-Qwen-Image-Lightning

Qwen-Image-Lightning: Speed up Qwen-Image model with distillation

Repository: https://github.com/ModelTC/LightX2V-Qwen-Image-Lightning
Canonical: https://ross.abutalabs.com/products/lightx2v-qwen-image-lightning
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-01-01T06:24:42+00:00

## Health v2 (maintenance only)
Score: 45/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 60, release rhythm 35, longevity 27
- inputs: {"age_days": 389, "days_push": 244, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1356, forks 48 (observed 2026-08-28T04:04:29.255400+00:00)

## What it is
Qwen-Image-Lightning provides distilled LoRA weights that accelerate the Qwen-Image text-to-image and image-editing models to 4-8 step inference while preserving complex text rendering. The repository releases safetensors checkpoints (fp32/bf16/fp8) and ComfyUI workflows for fast image generation.

## Use cases
- generate images with Qwen-Image in only 4 steps
- speed up text-to-image generation with distilled LoRA weights
- run fast Qwen-Image editing workflows in ComfyUI
- render complex text inside generated images quickly
- use fp8 fused checkpoints for efficient inference
- distill diffusion model inference steps for production image generation

## When to choose
- you already use Qwen-Image or Qwen-Image-Edit and want dramatically faster sampling
- you need few-step inference LoRAs compatible with ComfyUI or diffusers
- you want to preserve Qwen-Image's strong text rendering while cutting latency

## When to avoid
- you need a full training or inference framework rather than pretrained weights
- you use a different base image model than Qwen-Image
- you require maximum quality over speed and prefer the full multi-step base model

## Facets
- artifact type: dataset
- maturity: active
- function: machine-learning, deep-learning, image-processing, llm-inference
- domain: artificial-intelligence, machine-learning, image-processing, deep-learning
- platform: python, cross-platform
- tags: diffusion-models, model-distillation, text-to-image, image-editing, lora-weights, qwen-image, few-step-inference, safetensors, comfyui, gpu

## Member repositories
- ModelTC/LightX2V-Qwen-Image-Lightning (main) score 45

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:29.255400+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:41:52.822163+00:00, confidence not recorded.
  - readme: https://github.com/ModelTC/LightX2V-Qwen-Image-Lightning (fetched 2026-08-28T04:04:29.255400+00:00, sha b25012d6cb51)
- Data as of 2026-08-30T08:39:29.467469+00:00.