FantasyTalking
Alibaba AMAP
portrait + audio (+ optional motion prompt); Wan2.1-14B; PROVEN on RunPod
image-to-videotalking-head-animationaudio-drivenlip-sync
Apache-2.0 (adapter + Wan2.1 base)
Hallo3
Fudan
portrait + audio + scene prompt; long-clip, full-scene motion; H100-class
image-to-videotalking-head-animationaudio-drivenlip-sync
MIT (code); CogVideoX-5B weights, verify commercial
SadTalker
Tencent / XJTU
portrait + audio; fast warp/3DMM, runs on modest GPU or CPU
image-to-videotalking-head-animationaudio-drivenlip-sync
Apache-2.0 (code); bundles Wav2Lip NON-COMMERCIAL
EchoMimic
Ant Group
portrait + audio (+ optional landmarks/pose); SD1.5 / AnimateDiff
image-to-videotalking-head-animationaudio-drivenlip-sync
Apache-2.0 (code); weights research disclaimer, verify
Sonic
Tencent
portrait + audio; high quality, Stable-Video-Diffusion based
image-to-videotalking-head-animationaudio-drivenlip-sync
CC BY-NC-SA 4.0 (NON-COMMERCIAL); SVD base
AniPortrait
Tencent Games
portrait + audio; SD1.5 / AnimateDiff, 512x512, strong identity
image-to-videotalking-head-animationaudio-drivenlip-sync
Apache-2.0 (code); SD1.5 OpenRAIL-M base, verify
LivePortrait
Kuaishou
portrait + driving video; near real-time, warp-based, high quality
image-to-videoface-reenactmentvideo-drivenreal-time
MIT (code); swap out InsightFace detector to stay clean
FlashPortrait
Fudan + MSRA
portrait + driving video; Wan2.1-14B, infinite-length ID-preserving
image-to-videoface-reenactmentvideo-driven
Apache-2.0 (code + Wan2.1 base)
FaceFusion
Henry Ruhs
source face + target video; headless onnxruntime; pin hyperswap (not inswapper)
image-to-videoface-swap
OpenRAIL-AS (code); default hyperswap weights commercial-ok
GHOST (sber-swap)
Sber AI
source face + target video; one-shot, no per-face training
image-to-videoface-swap
Apache-2.0 (code + Sber weights)
SimSwap
SJTU
source face + target image/video; one-shot 224/512
image-to-videoface-swap
CC BY-NC 4.0 (NON-COMMERCIAL)
Deep-Live-Cam
hacksider
source face + target video, or real-time webcam swap
image-to-videoface-swapreal-time
AGPL-3.0 (code); inswapper_128 weights NON-COMMERCIAL
DeepFaceLab
iperov
cinematic quality; TRAINS per face-pair (hours-days), not on-demand
image-to-videoface-swap
GPL-3.0 (code); you train your own weights
Character swap / body
The WHOLE person - face, hair, body, clothes. Replace the person in a video, or animate a reference person from a photo + motion.
Wan-Animate
Alibaba Wan
replace the WHOLE person in a video with your character (Mix mode), keeps motion + expression; Wan2.x-based
image-to-videocharacter-swapvideo-driven
Apache-2.0 (Wan)
Wan VACE
Alibaba Wan
unified video edit: actor replacement, masked-region repaint, background swap; Wan2.1-based
image-to-videocharacter-swapvideo-driven
Apache-2.0 (Wan)
MimicMotion
Tencent
reference person photo + pose video -> full-body motion clip; SVD-based, ~16GB VRAM
image-to-videocharacter-swappose-driven
Apache-2.0 (code); Stability SVD base = NON-COMMERCIAL
MusePose (Animate Anyone)
Tencent Music / Lyra Lab
reference person + driving pose -> animated; MusePose = the open Animate-Anyone reimpl
image-to-videocharacter-swappose-driven
MIT (code); SD1.5 + AnimateDiff OpenRAIL-M base
UniAnimate
Alibaba + HUST
reference person + pose; efficient long-video (1 min) human animation
image-to-videocharacter-swappose-driven
NON-COMMERCIAL (research only); SD2.1 base
Animate-X
Ant Group
universal character animation (robust to stylized/anthropomorphic refs), pose-driven
image-to-videocharacter-swappose-driven
Apache-2.0 (code); SD2.1 OpenRAIL-M base
EchoMimic V2
Ant Group
semi-body avatar, audio + pose driven (talks and gestures); ~16GB VRAM
image-to-videocharacter-swappose-drivenaudio-driven
Apache-2.0 (code + weights); SD1.5 OpenRAIL-M base
Wan 2.2 i2v A14B
Alibaba Wan
image-to-video, ~5s, needs init_image
image-to-video
Apache-2.0
Wan-Alpha (transparent)
Wan + community DoRA
RGBA transparent video, ~5s
image-to-videotransparent-output
Apache-2.0
Wan 2.2 FLF (first-last frame)
Alibaba Wan
first + last frame conditioning, ~5s; needs init_image + last_frame
image-to-video
Apache-2.0
Stable Diffusion XL
Stability AI
1024x1024 (portrait/landscape ok)
text-to-imagepainterly
CreativeML OpenRAIL++
Juggernaut XL
RunDiffusion
1024x1536
text-to-imagephotorealisticpainterly
CreativeML OpenRAIL++
DreamShaper XL
Lykon
1024x1536
text-to-imagepainterly
CreativeML OpenRAIL++
RealVisXL V5
SG161222
1024x1536
text-to-imagephotorealistic
CreativeML OpenRAIL++
ZavyChroma XL
Zavy
1024x1536
text-to-imagepainterly
CreativeML OpenRAIL++
EpicRealism XL
epinikion
1024x1536
text-to-imagephotorealistic
CreativeML OpenRAIL++
Playground v2.5
Playground AI
1024x1024
text-to-imagepainterly
Playground v2.5 Community
Kolors
Kuaishou
1024x1024
text-to-imagepainterly
Apache-2.0
FLUX.1 dev
Black Forest Labs
1024x1536
text-to-imagepainterly
FLUX.1-dev Non-Commercial
Qwen-Image
Alibaba Qwen
1024x1536
text-to-imagepainterly
Apache-2.0
SANA 1.6B
NVIDIA
1024x1024 (fast)
text-to-image
NSCLv2
PixArt-Sigma
PixArt-alpha
1024x1024
text-to-imagepainterly
OpenRAIL++
AbsoluteReality (SD1.5)
Lykon
512x768
text-to-imagephotorealistic
CreativeML OpenRAIL-M
Realistic Vision V6 (SD1.5)
SG161222
512x768
text-to-imagephotorealistic
CreativeML OpenRAIL-M
DreamShaper 8 (SD1.5)
Lykon
512x768
text-to-imagepainterly
CreativeML OpenRAIL-M
Stable Diffusion 3.5 Large
Stability AI
1024x1024, gated weights
text-to-imagepainterly
Stability AI Community
Stable Image Ultra
Stability AI
hosted API, top-tier 1MP+
text-to-imagepainterly
Proprietary (Stability API)
HiDream-I1 Full
HiDream AI
1024x1024, needs a Llama text encoder
text-to-imagepainterly
MIT
HunyuanImage 3.0
Tencent
native multimodal, 1024x1024+
text-to-imagepainterly
Tencent Hunyuan Community
Z-Image
Tongyi Lab (Alibaba)
6B efficient, 1024x1024
text-to-imagepainterly
Apache-2.0
Z-Image Turbo
Tongyi Lab (Alibaba)
few-step distilled, fast 1024x1024
text-to-image
Apache-2.0
Imagen 4
Google
high-fidelity text-to-image
text-to-image
Proprietary (Vertex API)
Retro Diffusion (RD Pro)
Retro Diffusion
true pixel art, up to 256px
text-to-imagepixel-art
Commercial (RD API)
Pixel Art XL (SDXL LoRA)
nerijs
1024x1024 + kCentroid downscale
text-to-imagepixel-art
CreativeML OpenRAIL++
Pixel Art Redmond (SDXL)
ArtificialGuyBr
1024x1024
text-to-imagepixel-art
CreativeML OpenRAIL++
Nano Banana (Gemini 2.5 Flash Image)
Google
instruction image edit + generation; needs init_image for edits
instruction-editimage-to-image
Proprietary (Vertex / Gemini API)
GPT Image 2
OpenAI
top-tier image gen + instruction edit, up to 2K
instruction-editimage-to-image
Proprietary (OpenAI API)
Qwen-Image-Edit
Alibaba Qwen
instruction edit, needs init_image
instruction-editimage-to-image
Apache-2.0
FLUX.1 Kontext dev
Black Forest Labs
instruction edit, needs init_image
instruction-editimage-to-image
FLUX.1 Non-Commercial
Step1X-Edit
StepFun
instruction edit, needs init_image
instruction-editimage-to-image
Apache-2.0
InstructPix2Pix
Tim Brooks et al.
instruction edit, needs init_image
instruction-editimage-to-image
MIT
ACE-Step v1 3.5B
ACE Studio / StepFun
music, up to ~4 min
audio-generationmusic
Apache-2.0
Stable Audio Open
Stability AI
sound effects, up to ~47s
audio-generationsound-effects
Stability Community
TangoFlux
Declare Lab
sound effects, ~10s
audio-generationsound-effects
Non-Commercial (research)
Llama 3.2 3B
Meta
128k context, fast
text-generationchat
Llama 3.2 Community
Llama 3.3 70B
Meta
128k context, flagship open
text-generationchat
Llama 3.3 Community
Llama 3.1 8B
Meta
128k context, balanced
text-generationchat
Llama 3.1 Community
Qwen2.5 72B
Alibaba Qwen
128k context, top open general
text-generationchat
Qwen License
Qwen2.5 Coder 32B
Alibaba Qwen
code-specialized
text-generationcode-generation
Apache-2.0
DeepSeek-R1 70B
DeepSeek
reasoning / chain-of-thought
text-generationreasoning
MIT
Mistral Small 3 24B
Mistral AI
32k context
text-generationchat
Apache-2.0
Gemma 3 4B (vision)
Google
multimodal text + vision (i2text); self-hosted vLLM on RunPod
text-generationchatimage-understanding
Gemma
Phi-4 14B
Microsoft
small but strong
text-generationchat
MIT
Llama 3.2 Vision 11B
Meta
image understanding (i2text); self-hosted vLLM on RunPod
text-generationimage-understanding
Llama 3.2 Community
Qwen2.5-VL 7B
Alibaba Qwen
image + document understanding (i2text); self-hosted vLLM on RunPod
text-generationimage-understanding
Apache-2.0
LLaVA 1.6 (Mistral 7B)
LLaVA
lightweight vision-language (i2text); self-hosted vLLM on RunPod
text-generationimage-understanding
Apache-2.0
Claude Opus 4.8
Anthropic
frontier reasoning + vision, 200k+ context
text-generationchatreasoningimage-understanding
Proprietary (Anthropic API)
Claude Fable 5
Anthropic
frontier creative writing
text-generationchatcreative-writing
Proprietary (Anthropic API)
Claude Sonnet 5
Anthropic
balanced frontier, fast
text-generationchatreasoningimage-understanding
Proprietary (Anthropic API)
Claude Haiku 4.5
Anthropic
fast + cheap frontier
text-generationchat
Proprietary (Anthropic API)
GPT-4o
OpenAI
multimodal frontier + vision (text + image-to-text)
text-generationchatimage-understanding
Proprietary (OpenAI API)
Gemini 2.5 Pro
Google
long-context multimodal, 1M tokens
text-generationchatimage-understanding
Proprietary (Google API)