AlayaLab/Evoke-Turbo

Alaya-EVOKE: From Linear-Scaling Supervision to Endless World

Alaya-EVOKE-Turbo is a controllability-optimized version of Alaya-EVOKE, focusing on camera-motion following and scene/subject instruction adherence. It uses the same architecture and inference workflow: 3 steps, no CFG, 384 × 640 at 24 fps.

Contents

Merged Turbo transformer weights, stored in FP32 to match Evoke. Download the shared base components and depth backend separately:

models/
├── evoke-base/                 # VAE / text encoder / tokenizer / scheduler; from Evoke
├── evoke-turbo/transformer/    # this release
└── ViGeo1.1/                  # required depth backend

Usage

git clone https://github.com/AlayaLab/Evoke && cd Evoke
pip install -r requirements.txt

hf download AlayaLab/Evoke --include "evoke-base/*" --local-dir models
hf download AlayaLab/Evoke-Turbo --local-dir models/evoke-turbo/transformer
hf download pkqbajng/ViGeo --local-dir models/ViGeo1.1   # REQUIRED depth backend

TRANSFORMER_PATH=models/evoke-turbo MODE=t2v NUM_CHUNKS=20 \
bash scripts/inference/infer_post_distill.sh

See Evoke for camera control, other inference modes, and long rollouts.

WBench

Results on WBench (158 cases; Interaction: Navigation only). Bold marks the best score in each column.

Model	Video Quality	Setting	Navigation	Consistency	Physical	Avg
Evoke	82.7900	83.7600	78.6300	86.8700	72.0550	80.8210
Evoke-Turbo	81.8914	82.0518	83.8978	88.1469	74.0133	82.0003

Turbo uses seed-44 results selected from six prompt variants; Evoke uses published reference scores, with different prompts and poses.

Notes

See Evoke for base-component provenance and dependency licenses; external components retain their own licenses.

Citation

@article{evoke2026,
  title   = {Alaya-EVOKE: From Linear-Scaling Supervision to Endless World},
  author  = {Yin, Yuanyang and Wang, Gongxuan and Zhan, Yifan and
             Li, Chuanhao and Zhang, Kaipeng and Zhao, Feng},
  journal = {arXiv preprint arXiv:2608.13546},
  year    = {2026},
}