OmniRT supported models¶
本页由 scripts/generate_models_doc.py 从注册表自动生成。
修改请动 registry，不要直接改这个文件。
Models are organized by the digital-human production chain rather than by a generic multimodal taxonomy.
Core avatar rendering¶
| Registry id | Task | Tier | Maturity | Realtime | Summary |
|---|---|---|---|---|---|
longcat-video-avatar-1.5 |
audio2video |
core | beta | no | LongCat-Video-Avatar 1.5 audio-driven avatar generation via external CUDA or Ascend checkouts. |
soulx-flashhead-1.3b |
audio2video |
core | beta | no | SoulX-FlashHead low-latency talking-head generation via image plus audio. |
soulx-flashtalk-14b |
audio2video |
core | beta | yes | SoulX-FlashTalk talking-head avatar generation via image plus audio on CUDA or Ascend. |
soulx-liveact-14b |
audio2video |
core | beta | no | SoulX-LiveAct long-form audio-driven avatar video generation on Ascend. |
Voice generation¶
| Registry id | Task | Tier | Maturity | Realtime | Summary |
|---|---|---|---|---|---|
cosyvoice3-triton-trtllm |
text2audio |
core | beta | no | CosyVoice3 text-to-audio generation through a Triton-compatible service endpoint. |
indextts |
text2audio |
adjacent | beta | yes | IndexTTS-2 resident text-to-audio service for OpenTalking TTS, with segment streaming and experimental token-window streaming through serve-text2audio. |
soulx-podcast-1.7b |
text2audio |
core | beta | no | SoulX-Podcast text-to-audio generation through a FastAPI service endpoint. |
Voice understanding roadmap¶
| Registry id | Task | Tier | Maturity | Realtime | Summary |
|---|---|---|---|---|---|
sensevoice-small |
audio2text |
core | beta | no | SenseVoice offline audio transcription for digital-human voice understanding. |
Avatar asset generation¶
| Registry id | Task | Tier | Maturity | Realtime | Summary |
|---|---|---|---|---|---|
chronoedit |
edit |
experimental | beta | no | ChronoEdit physically-consistent image editing pipeline. |
flux-canny |
edit |
adjacent | beta | no | Flux canny-guided structured image generation pipeline. |
flux-depth |
edit |
adjacent | beta | no | Flux depth-guided structured image generation pipeline. |
flux-kontext |
edit |
adjacent | beta | no | Flux Kontext image editing pipeline. |
qwen-image-edit |
edit |
adjacent | beta | no | Qwen-Image single-image editing pipeline. |
qwen-image-edit-plus |
edit |
adjacent | beta | no | Qwen-Image multi-reference editing pipeline. |
qwen-image-layered |
edit |
adjacent | beta | no | Qwen-Image layered decomposition pipeline. |
sdxl-refiner-1.0 |
image2image |
adjacent | beta | no | SDXL refiner image-to-image pipeline for second-stage refinement passes. |
flux-fill |
inpaint |
adjacent | beta | no | Flux Fill inpainting and outpainting pipeline. |
bria-3.2 |
text2image |
experimental | beta | no | Bria 3.2 commercial-ready text-to-image pipeline. |
flux-dev |
text2image |
adjacent | stable | no | Flux 1 dev text-to-image pipeline. |
flux-schnell |
text2image |
experimental | stable | no | Flux 1 schnell low-step text-to-image pipeline. |
flux2.dev |
text2image |
adjacent | beta | no | Flux 2 dev text-to-image pipeline. |
glm-image |
text2image |
experimental | beta | no | GLM-Image instruction-following text-to-image pipeline. |
hidream-i1 |
text2image |
experimental | beta | no | HiDream-I1 modern text-to-image pipeline. |
hunyuan-image-2.1 |
text2image |
experimental | beta | no | Hunyuan Image 2.1 text-to-image pipeline. |
kolors |
text2image |
experimental | beta | no | Kolors multilingual text-to-image pipeline. |
lumina-t2x |
text2image |
experimental | beta | no | Lumina-T2X text-to-image pipeline via the LuminaPipeline runtime. |
omnigen |
text2image |
experimental | beta | no | OmniGen text-to-image generation path. |
ovis-image |
text2image |
experimental | beta | no | Ovis-Image text-heavy generation pipeline. |
pixart-sigma |
text2image |
experimental | beta | no | PixArt-Sigma high-resolution text-to-image pipeline. |
qwen-image |
text2image |
adjacent | beta | no | Qwen-Image multilingual text-to-image pipeline. |
sana-1.6b |
text2image |
experimental | beta | no | Sana 1.6B efficient text-to-image pipeline. |
sd15 |
text2image |
experimental | beta | no | Stable Diffusion 1.5 baseline text-to-image pipeline. |
sd21 |
text2image |
experimental | beta | no | Stable Diffusion 2.1 text-to-image pipeline. |
sd3-medium |
text2image |
experimental | beta | no | Stable Diffusion 3 Medium text-to-image pipeline. |
sd3.5-large |
text2image |
experimental | beta | no | Stable Diffusion 3.5 Large text-to-image pipeline. |
sd3.5-large-turbo |
text2image |
experimental | beta | no | Stable Diffusion 3.5 Large Turbo text-to-image pipeline. |
sdxl-base-1.0 |
text2image |
adjacent | stable | no | SDXL base text-to-image pipeline with LoRA support. |
sdxl-turbo |
text2image |
experimental | beta | no | SDXL Turbo low-latency text-to-image pipeline. |
Video and idle assets¶
| Registry id | Task | Tier | Maturity | Realtime | Summary |
|---|---|---|---|---|---|
helios-i2v |
image2video |
experimental | beta | no | Helios image-to-video pipeline. |
hunyuan-video-1.5-i2v |
image2video |
experimental | beta | no | HunyuanVideo 1.5 image-to-video pipeline. |
kandinsky5-i2v |
image2video |
experimental | beta | no | Kandinsky 5 Pro image-to-video pipeline. |
ltx2-i2v |
image2video |
experimental | beta | no | LTX image-to-video pipeline. |
svd |
image2video |
adjacent | stable | no | Stable Video Diffusion base image-to-video pipeline. |
svd-xt |
image2video |
adjacent | stable | no | Stable Video Diffusion XT image-to-video pipeline. |
wan2.1-i2v-14b |
image2video |
experimental | beta | no | Wan 2.1 image-to-video pipeline. |
wan2.2-i2v-14b |
image2video |
adjacent | beta | no | Wan 2.2 image-to-video pipeline. |
animate-diff-sdxl |
text2video |
adjacent | beta | no | AnimateDiff SDXL text-to-video pipeline. |
cogvideox-2b |
text2video |
experimental | beta | no | CogVideoX 2B text-to-video pipeline. |
cogvideox-5b |
text2video |
experimental | beta | no | CogVideoX 5B text-to-video pipeline. |
helios-t2v |
text2video |
experimental | beta | no | Helios text-to-video pipeline. |
hunyuan-video |
text2video |
experimental | beta | no | HunyuanVideo text-to-video pipeline. |
hunyuan-video-1.5-t2v |
text2video |
experimental | beta | no | HunyuanVideo 1.5 text-to-video pipeline. |
kandinsky5-t2v |
text2video |
experimental | beta | no | Kandinsky 5 Pro text-to-video pipeline. |
ltx-video |
text2video |
experimental | beta | no | LTX-Video text-to-video pipeline. |
mochi |
text2video |
experimental | beta | no | Mochi text-to-video pipeline. |
sana-video |
text2video |
experimental | beta | no | Sana-Video efficient text-to-video pipeline. |
skyreels-v2 |
text2video |
experimental | beta | no | SkyReels-V2 text-to-video pipeline. |
wan2.1-t2v-14b |
text2video |
experimental | beta | no | Wan 2.1 text-to-video pipeline. |
wan2.2-t2v-14b |
text2video |
adjacent | beta | no | Wan 2.2 text-to-video pipeline. |
Aliases¶
| Alias | Canonical id |
|---|---|
flux2-dev |
flux2.dev |