Krea 2 Identity Edit
ð§Ž
250
Identity-preserving instruction image editing on Krea 2
Community Spaces
Apps started here then claimed by the ð authors
-
-
CharacterSheet LoRA Demo
ð§198Multi-view character sheet from one image (FLUX.2 LoRA)
-
Mage-Flow
ðĻ126Efficient native-resolution image generation and editing
-
Hum to Song
ðĪ118Hum a melody, get a finished song
-
Krea 2 Outpaint
ðž114Extend images into larger canvases with Krea 2 outpaint
-
Mage-VL
ð108Codec-native video & image understanding with Mage-VL 4B
-
Breeze TTS 2
ð97Bilingual TTS with voice design, cloning, and direction
-
ABot-World Interactive
ð96Live interactive world rollout from an image
-
JEV-9B â typed decisions
â92Calibrated typed decisions + System 2 reasoning
-
Krea 2 Turbo BBox
ð§Đ56Layout-controlled comics with a draggable bbox canvas
-
Ming-Image 0.1 Design
ðĻ56Text-to-image model for UI, infographics and poster design
-
LTX-Best-Face-ID
ðŽ53Distilled LTX-2.3 identity video from a reference photo
-
NaviDC-OCR
ð51Parse digital & photographed documents into Markdown
-
Nemotron-Labs-Audex
ð§47Unified audio-text intelligence
-
CADENA Stepwise CAD
ð46Turn a 3D mesh into a parametric CadQuery program
-
LTX Ripple
ðŽ45Edit one frame, ripple the change through the whole video
-
ABot-Recon Streaming 3D
ð41Streaming 3D reconstruction from video with local context
-
VibeVoice ASR BitNet
ð39Multilingual CPU-only ASR with a 1.58-bit BitNet decoder
-
SpotSound Temporal Grounding
ð35Find when a described sound happens in a recording
-
MOSS-VL-Realtime
ð33Realtime VLM for image and video understanding
-
MiniMax-H3 360° Orbit LoRA
ð28One photo into a frozen-instant 360° camera orbit
-
A.X K2 Raon-Speech 21B-A3B
ðĢ25Bilingual EN/KO speech LM - transcribe, ask, and speak
-
Audio8 TTS Preview 0.1B
ðĢ25Zero-shot voice cloning TTS with Audio8 0.1B
-
CrisperWhisper 2.0 Controllable ASR
ð22Verbatim + intended transcripts with word-level timing
-
SenseNova U1 Infographic V3
ð18Infographic generation & editing with SenseNova-U1 V3
-
StuPASE Speech Enhancement
ð18Studio-quality generative speech enhancement
-
Video-ORA-9B
ðŊ17Explore all seven Video-ORA task families
-
SCoPE Camera Controlled Video
ðĨ16Steer a 3D camera path through any still image
-
Paradee
ðĶ198M-param English TTS distilled from Kokoro-82M
-
WordVoice Word-Level Controllable TTS
ðĢ14Zero-shot TTS with explicit word-level prosody control
-
Aerial Imagery Object Detection
ð13Aerial object detection - YOLO Models
-
ClinFusion Medical VLM
ðĨ13ClinFusion medical multimodal LLM for 2D images
-
LTX 2.3 Pixel Spatial Upscaler
ð13Generative 4x video upscaling with LTX-2.3 IC-LoRA
-
LTX-2.3 Clean Plate (Remove People from Video)
ðŽ13Remove people & vehicles from video, keep the background
-
Supra2-IMG
ðž13104M-param text-to-image DiT, trained from scratch
-
PhysBrain 1.5-8B
ðĶū12Embodied spatial reasoning, pointing and arm traces
-
Julia-1 Decision Demo
ð§12Typed decision model: state in, choices + probabilities
-
PlagueKind MiniMax H3
ðĶ13The PlagueKind V1.5 workflow for MiniMax-H3 video + audio
-
MonkeyOCRv2 Document Parser
ð11Parse document images into structured Markdown
-
QueenVIS Video Instance Segmentation
ð10Segment and track object instances in video with QueenVIS
-
Krea-2 Depth ControlNet
ð10Depth-controlled image generation with Krea-2 Turbo
-
MiLMMT-46 Translation (12B)
ð10Multilingual translation across 46 languages with MiLMMT
-
Qwen-Image-2.1 Object Remover Bbox Turbo
ð10Box an object and let Qwen-Image-2.1 erase it.
-
Sopro v2 Turbo TTS
ðĢ10Zero-shot voice cloning TTS with 120M params
-
Zing 0.5 World Model
ðŪ9Walk through a world Zing-0.5 hallucinates live around you
-
S1-mini Transcript Cleanup
â9Clean and normalize raw ASR transcripts with S1-mini
-
MobileWan T2V
ðŽ9MobileWan text-to-video generation (Qualcomm AI Research)
-
Krea 2 Style Reference
ðĻ8Style-guided image generation with Krea 2 Turbo
-
MLTNT Militant Roots Reggae (YuE2)
ðĨ8Militant roots reggae LoRAs for YuE2-3B
-
WorldCrafter
ð8Camera-controlled video world model with 3D-aware memory
-
Krea 2 Turbo HD
ðĻ7Krea 2 Turbo HD text-to-image with enhanced VAE
-
Glint-2 Effort Explorer
ð71.06M-param pure-loop transformer with six effort levels
-
MiniMax-H3 Spatial Physics LoRA
ðŽ7Spatial physics video generation with MiniMax-H3 LoRA
-
Mr. Pong
ð7Play table tennis against a PPO agent
-
LTX-2.5 BBox Control
ðĶ7Animate bounding boxes into video, one prompt per box
-
EdgeCrafter
ðŠķ7Detection, segmentation & pose with compact edge ViTs
-
Flux2-Klein-9B-True-V3 Edit
ðĻ6FLUX.2 Klein 9B fine-tune for image generation and editing
-
HPD-Parsing
ð6Hierarchical parallel document parsing with a 1B VLM
-
Tafsut Forecast
ð6Zero-shot probabilistic time-series forecasting
-
AudioRubrics
ð§6Audio reasoning with evolving rubric rewards
-
ControlFoley Video-to-Audio
ðŽ6Unified controllable video-to-audio Foley generation
-
GLiNER2 Guardrails PII Multi
ð6Multilingual PII detection & LLM safety moderation
-
LTX-2.3 Relight (Video Relighting IC-LoRA)
ðĄ6Relight exterior video clips with a light-direction ball
-
Index-Echo-S2ST-9B
ð7Voice-preserving speech dubbing across zh/en/es/ja
-
Reproduction: Matroid Algorithms Under Size-Sensitive Independence Oracles
ð5Native size-sensitive matroid audit
-
Chatterbox-Flash TTS
ð5Block-diffusion zero-shot text-to-speech by Resemble AI
-
Supra-A2A-Nano-Exp
ð5Tiny any-to-any multimodal GPT (text + image) prototype demo
-
OpenCaption-4B-VL-SFT
ð5Dense fine-grained image captioning with Qwen3-VL-4B
-
Kokoro Inno Clone Tuner
ð5Zero-shot voice tuning for Kokoro-82M from a reference clip
-
SupraElegans-500K Explorer
ðŠą5Watch a 574K-param non-Transformer LM think
-
XGEN-JING Demo
ðŪ5Explore, interact and converse in first-person video
-
Confucius4-R2T2
ð5Offline ASR with Confucius4-R2T2
-
zen-image-edit
ðŠ·5Qwen-Image-2.1 editing with a 0.8B text encoder
-
DMAD â 4-step MiniMax-H3 student
ðŽ54-step text-to-audio-video generation, distilled from H3
-
NK2E Krea 2 Image Editor
ðĻ4Edit images or generate from Canny edges with NK2E on Krea 2
-
ViT-Up Feature Upsampler
ðž4DINOv3 feature upsampling with ViT-Up
-
LightOn-rerank-PW
âĻ4LightOn multimodal document reranker with pointwise scoring
-
LTX-2.3 Product Ad Style
ðŽ4Cinematic product commercial style video LoRA for LTX-2.3
-
Jolia
ðŦ4Zero-shot CT findings with the Jolia 3D CT foundation model
-
Play Charles the Chess Bot
â4Play chess against a 32.5M-param search-free policy
-
LittleLearner 5B Chatty
â4Chat with a 5B model that only studied grades K-5
-
TinyCast Forecaster
ð4TinyCast zero-shot probabilistic time-series forecasting
-
Aero Realtime 4B
ðš4Proactive real-time commentary on audio-video streams
-
saga-2 Danish ASR
ð4Danish ASR: RoPE Whisper encoder + Qwen3 decoder
-
Jina Reranker v3.5 Demo
ð4Listwise document reranker with jina-reranker-v3.5
-
Semancer 12B
ðŪ4Occult philosophy chat LLM fine-tuned on Gemma 4 12B
-
Audio8-ASR-0.1B
ð4Compact multilingual ASR model (324M params)
-
Arithmetic-SLM
ð§Ū4Tiny 31.7M SLM that solves arithmetic expressions
-
ASASR Super-Resolution
ð4Faithful x4 image super-resolution via FLUX.1-dev dual-LoRA
-
Krea 2 HDR LoRA (LogC4)
ð3Generate scene-linear HDR images with Krea 2 + LogC4 LoRA
-
Scrappy Voice TTS
ðĢ3Four voices in one 10M-param CPU model, with blending
-
Inline Skin LoRA â Krea 2
ðĻ3Photorealistic skin-texture LoRA for Krea 2 Turbo
-
OpenDDE Structure Prediction
ð§Ž3Predict 3D protein structures with OpenDDE
-
Mr. Balance
ð―3An RL plate that balances 21 objects in MuJoCo
-
EverAnimate
ðš3Minute-scale human animation with Wan2.2-Animate + LoRA
-
LULA-1.1 Protein-Ligand Binding Scorer
ð§Ž3Sequence-only protein-ligand binding scorer (LULA-1.1)
-
TontaubeV1 TTS
ð3Multilingual text-to-speech with zero-shot voice cloning
-
SAGE Visual Place Recognition
ðš3SAGE retrieves matching place images from a gallery
-
WATERec WordArt Recognition
ðĻ3Recognize text from WordArt / artistic scene text images
-
Arabic Speech Tashkeel
ð3Arabic speech to fully-diacritized text (tashkeel)
-
X2-Turn-4B
ðĢ3Bilingual streaming ASR with turn-taking prediction
-
MOSS-VL-Instruct-0708
ð§3Image and video understanding with MOSS-VL multimodal model
-
Text-to-CT Generation
ðŦ3Synthesize a 3D chest CT from a radiology report
-
PhenoSeq
ð§Ŧ3Cell Painting images to single-cell transcriptomes
-
Cua-S1-4B GUI Decision Model
ðą3One-pass GUI (element, action) decision scorer
-
Eikos-4B Typed Decisions
â3Single-pass typed decisions with calibrated probabilities
-
SupraTTS-0.1-Beta
ðĢ3From-scratch Glow-TTS + HiFi-GAN English voice
-
Samudra 2
ð3Roll a neural ocean emulator forward against GFDL-OM4
-
LayerRecall Long Video
ðŽ3Multi-shot video with LayerRecall long-horizon memory
-
DomainShuttle Subject Video
ðŽ2Subject-driven text-to-video from reference images (Wan2.2)
-
MolParser-Mobile
ð§Š2Recognize molecular structure images as E-SMILES
-
AlienLM
ðļ2AlienLM privacy layer for black-box LLMs
-
Translate-15L
ð2Compact 60M T5 translator for 15 languages
-
UniAR
ðĻ2Unified AR model for image understanding & generation
-
SauerkrautLM-LFM2.5-GLiNER
ðĨŽ2Zero-shot multilingual NER with GLiNER on LFM2.5-350M
-
Tiron Meeting Transcriber
ð2Multi-speaker meeting transcription with diarization
-
Histoformer All-Weather Restoration
ð§2Remove adverse weather effects (Histoformer, ECCV 2024)
-
MiLMMT-46 Translation (4B)
ð2Multilingual translation across 46 languages with MiLMMT
-
CoinVE-Edit
ðŽ2Compositional instruction-guided video editing
-
ForgeWM
ðŪ2Few-step action-conditioned Minecraft world model
-
LATO.2
ðš2Rebuild any 3D mesh as a clean artist-style mesh
-
Chipoint v2 Geolocation
ð2Predict where a photo was taken â full chipoint-2, both arms
-
DoctoBERT Fill-Mask
ðĐš2Fill-mask demo for a French medical ModernBERT
-
SANA-Video 2.0 5B 720p · 4-Step Preview
ðŽ2Four-step 720p text-to-video preview with SANA-Video 2.0
-
Supertron3 0.8B
ðą2GUI grounding & tool calling with Supertron3-0.8B
-
Krea 2 Lineart Edit
â2Paint inside your line art with Krea 2
-
Code-as-World VL Demo
ð2Video-based quantitative physical reasoning VLM
-
grug-v2-9b
ðĶ2Multimodal reasoning VLM with Grug-style thinking
-
JoyAI-Echo x LTX-2.3 Multi-Shot Video
ðŽ2Multi-shot narrated video with cross-shot memory
-
Lumma 0.6B Tool Playground
ð2Define JSON tools and watch Lumma-0.6B generate tool calls
-
Kev-4B Decision Model
ð2Typed questions in, calibrated probabilities out
-
LLaDA-UI
ðą2GUI grounding & agent with a block-diffusion VLM
-
Index-Translate-9B
ð2150-language translation with instruction following
-
UrbanAnonymizer
ðĩ2Mask faces and license plates in street-level images
-
FarmerChat Crop Diagnosis
ðŋ2Configurable multi-stage crop disease and pest diagnosis
-
Ear â Spoken Language ID
ð2Detect the spoken language of a recording, 99 languages
-
ImIR: All-in-One Image Restoration
ðĪ2Prompt-free all-in-one restoration (rain/haze/blur/noise)
-
Gero-4B
â2Calibrated probabilities for decisions, no generated text
-
Whisper Karakalpak ASR
ð1Karakalpak speech-to-text with fine-tuned Whisper Medium
-
Miril-DroneVLM-2B-2 Demo
ð1Speak to a drone using an image, video, text, or voice.
-
EpistemeAI Reasoning-Medical-27B
ðĨ1Medical reasoning VLM based on Qwen3.6-27B
-
OmniAgent-RL
ðŽ1Agentic video understanding with active perception loop
-
BioMatrix-4B-SFT
ð§Ž1Multimodal bio foundation model for molecules & proteins
-
MolmoMotion
ðŊ1Forecast 3D point trajectories from video and language
-
LabVLA
ð§Š1VLA model for scientific laboratory robotics
-
DI2FIX Enhancer
âĻ1Distractor-free 2D enhancer for radiance field renderings
-
WSA-Base Robot Control
ðĪ13D-centric world-spatial-action robot policy demo
-
JoyAI Image Edit Plus
ðĻ1Multi-image instruction-guided image editing
-
SHINE Context to LoRA
ð1Encode a context into a LoRA and chat with Qwen3-8B
-
Parti Hardware Review
ð§1Draft hardware plans, review sources, and record test notes
-
Ava Persian TTS
ðĢ1Persian TTS demo using Ava-82M (Kokoro-based)
-
MimiLens Watermark Detector
ð1Detect watermarks, signatures, and artifacts in images
-
BarunAction-35M
ð§1Proposal-only typed personal actions with a 35M-param model.
-
ShotPlan Cinematic T2V
ðŽ1Multi-shot cinematic text-to-video generation
-
MBA Business Ideation Agent
ðĄ1Generate market-grounded business ideas from images with MBA
-
VideoSearch-R1
ðŽ1Video verification & temporal grounding with VideoSearch-R1
-
MiLMMT-46 Translation (1B)
ð1Multilingual translation across 46 languages with MiLMMT
-
Hayai OCR v2
ð1Compact OCR model for Japanese, Chinese & English text
-
TAU-Agent Traffic Anomaly Understanding
ðĶ1Ask questions about traffic anomalies in video
-
SpeakoFlow Mini
ðĪ1Dictation cleanup for transcribed speech (0.8B GGUF)
-
TASKER Keyframe Extractor
ð1VLM-guided tree-search keyframe extraction from videos
-
RenderFormer V2
ðŦ1Neural rendering of 3D scenes with a transformer
-
METEOR Multi-Task Driving Perception
ð18-camera multi-task driving perception, 12 heads at once
-
Baberu OCR
ð1Multilingual manga speech-bubble OCR (JA/ZH/EN)
-
LimiX-2 Tabular In-Context Learning
ð1In-context classification, regression & imputation on tables
-
Stop-Motion Consistency LoRA (Qwen-Image 2.1)
ð1Next-frame stop-motion generation with Qwen-Image 2.1 + LoRA
-
FACT World-Action Model
ðĪ1Counterfactual robot futures from a world-action model
-
FreeFlow Optical Flow
ð1Estimate optical flow between two frames with FreeFlow
-
FLUX.2 Klein Alpha â RGBA
â1RGBA object cut-out and object removal
-
Index-Echo-S2ST-2B
ð1Voice-preserving speech dubbing across zh/en/es/ja
-
GRVL Raspy Rock-Soul Vocals (YuE2)
ðŠ1Raspy female rock-soul songs with YuE2 + GRVL LoRAs
-
Edda v0.1 Danish ASR
ð1Danish speech-to-text with Edda v0.1 Whisper model
-
Rene-1 â typed decisions
â1Typed questions in, calibrated probabilities out
-
VoxMind Agentic Spoken Dialogue
ðĢ1Speech-in, reasoning, tool calls, speech-out in one model
-
EMA-TTS Turkish
ðĢ1Turkish TTS via a 65M flow-matching DiT (48 kHz)
-
Index-Echo-S2TT-2B
ðŽ1Chinese speech or video to bilingual timestamped subtitles
-
BuzzASR Persian
ð1Persian speech recognition with BuzzASR (Whisper-large-v3)
-
K2-Type-0.9B
â1Typed decisions (yes/no, choice, score) in one pass
-
H3 Keyframe Animation
â1Hand-drawn keyframe animation LoRAs for MiniMax-H3
-
Spotlight VM
ðĶ1A tiny transformer that runs Python programs
-
UNCHA Hyperbolic VLM
ðŪHyperbolic vision-language zero-shot classification
-
MeanFlowNFT Few-Step Generation
âĄ4-step RL text-to-image with MeanFlowNFT on SD3.5-Medium
-
LocalSubs EN-ZH-TW Subtitle Translation
ðŽEN subtitle cues to Taiwan Traditional Chinese (0.6B model).
-
Temporal-MoEs-RoBERTa Emotional State-Change Forecaster
ð§Forecast Valence & Arousal state change from posts
-
Graph-PrefLexOR Scientific Reasoning
ð§Graph-native LLM scientific reasoning with graph viz
-
MOSS-TTS-PNY
ðSpeaker-conditioned TTS with emotion & energy control
-
MrFlow Fast Diffusion
âĄMrFlow training-free diffusion acceleration demo
-
LoomVideo
ðŽUnified multimodal video generation and editing (5B)
-
InterleaveThinker
ðĪMulti-agent interleaved text-image generation pipeline
-
Perceive-to-Reason
ðP2R fine-grained visual reasoning with Qwen3-VL
-
HOMA
ðŽHuman-object interaction video from image+audio+text
-
AffectFlow-DINO
ðFacial affect estimation with uncertainty via rectified flow
-
TimEE In-Context Time Series Classification
ðClassify time series in-context with TimEE foundation model
-
LTX CrossView Warp
ðĨNovel camera viewpoint from a video via depth-warp IC-LoRA
-
EchoVLM Ultrasound Demo
ðĐšUltrasound image understanding VLM with MoE architecture
-
SamplingTAR Defense
ðĄTraining-free typographic attack defense for CLIP
-
LongE2V Event-Video Reconstruction
ðŽReconstruct video from event-camera voxel grids via LongE2V
-
Lung Ultrasound Video Classification
ðŦ4-class lung ultrasound video classifier with attention
-
DAV-Det AIGC Audio-Video Detector
ðĨDetect AI-generated audio-visual content (DAV-Det)
-
SpatialAxiom-9B
ð§Spatial reasoning VLM for 3D relations and perspective
-
PEC Breast MRI VCE
ðĐŧSynthesize contrast-enhanced breast MRI from pre-contrast
-
Korean Toxicity Deobfuscation KOTOX
ð§žDeobfuscate and detoxify Korean text with KOTOX
-
Translate Gemma 4 Sub E4B
ðContext-aware multilingual subtitle translation with Gemma 4
-
MiniMax-H3 Wushu Action LoRA
ðĨMartial-arts action video with a MiniMax-H3 LoRA
-
JoyAI-Echo LTX-2.5 echoVid
ðŽJoyAI-Echo on LTX-2.5, few-step video with audio
-
CW-BASS v2 Segmentation
ðŊDINOv2-based semi-supervised semantic segmentation
-
Guardian Failure Detection
ðĄDetect robot planning & execution failures with a VLM
-
DEAR AI Image Detection
ðDetect AI-generated images with DEAR (Dissect and Prune)
-
Iwin Transformer Image Classification
ðŠIwin Transformer for ImageNet image classification
-
EXAONE Tabular
ðIn-context tabular prediction from a CSV
-
Griot-MT Translation
ðTranslate between French and 18 African languages
-
Causilo
ðIn-context tabular prediction from a CSV
-
EventMemAgent Video QA
ðOnline video QA with event-centric memory + tool use
-
AlayaWorld
ðCamera-controlled interactive video world model
-
MiniMax Music 3 Particle Sliders
ðTurn one musical property up with a LoRA slider
-
ASMR Trigger & Audio · MiniMax-H3 LoRA
ð§Whispered ASMR audio-video LoRA for MiniMax-H3
-
HiMoE-VLA Action Prediction
ðĪPredict robot action chunks from an image + instruction
-
EM3M-Gen
ðŽGenerate electron micrographs from text prompts
-
V-Retrver
ðAgentic multimodal reranking with visual tool use
-
CamDistill Camera Motion Demo
ðĨTemporally grounded camera-motion analysis for video clips
-
AgentJev 0.6B
ðState in, a probability distribution out. Nothing decoded.
-
rakedoc-nano
ðDocument page to markdown with rakedoc-nano (1.2B)
-
OmniJev-4B
ðŊTyped questions on an image, answered with probabilities
-
Azerbaijani OCR â CRNN + CTC
ðRead Azerbaijani text lines and whole page scans
-
QWWL / DRKSF Qawwali & Sufi LoRAs (YuE2)
ðQawwali and dark Sufi fusion LoRAs for YuE2-3B
-
open-jev â typed decisions
ð§Answer typed questions about a text in one forward pass
-
Yug â Zero-Shot Probabilistic Time-Series Forecasting
ðZero-shot quantile forecasts for any time series
-
Bonsai Image 4B LiteRT
ðŠīTernary diffusion transformer on CPU via LiteRT
-
Nemotron ASR Streaming â Persian
ðStreaming Persian (Farsi) speech recognition
-
SilvermoonMix Anima Evolved
ðAnime text-to-image with SilvermoonMix Anima 2.9B
-
Decision-2.0-Lux-9B
âTyped decisions in one forward pass â probability per answer
-
MGFlow
âĄSOTA one-step 512Ã512 generation with FLUX.2 [klein] 4B