LocalOps

NVIDIA humanoid foundation model

Shadows human motion to robot

Transformer image restoration

Image super-resolution

Ultra-fast local TTS

Latest YOLO model

Fast English embeddings

Face restoration

Minimal ASR model

Large YOLO variant

Latest YOLO object detection flagship

Robust face restoration

Object detection flagship

Next-gen background removal, video support

Highest quality lightweight TTS

Compact ASR

Tiny real-time model (0.1s latency)

Open source robot manipulation policy

Zero-shot TTS/Voice conversion

Fast variational inference TTS

Multilingual OCR

Human pose estimation

Image prompt adapter

Face ID preservation

Commercial background removal

Balanced embeddings

Document understanding

Smallest practical language model

Long context embeddings (8192)

Unified speech/text model

Fast and accurate

Fast expressive TTS

Background removal

Fast ASR optimized for resource constrained

High-res Gaussian splatting

Rhymes AI open video model

OCR-free document understanding

Easy multilingual OCR

Multi-person pose detection

Bilateral reference high-resolution segmentation

Video segmentation

Segment Anything in images and video, improved accuracy

Fast ASR

Real-time streaming voice cloning

Ultra-compact on-device model

Zero-shot voice cloning

Alibaba's multilingual TTS

Multilingual fast TTS

Fast music generation

Self-supervised ASR

Mixed bread AI embeddings

English embeddings

General text embeddings

Contrastive embeddings

Monocular depth estimation

High-quality monocular depth estimation

Metric depth estimation

Nanoscale efficient model

8k context, modern architecture

Modernized BERT for efficient retrieval, 8K context

Turn any SD model into video

SOTA commercial-friendly embedding

Efficient TTS

Google improved CLIP

Efficient embeddings

Low-latency high-performance TTS

Vision-language alignment

Smallest Qwen variant

Streaming speech synthesis foundation

Fine-grained emotional control

High quality voice cloning

Fast image to 3D mesh

Single image to 3D

Artist-created mesh alignment

Natural voice cloning TTS

Ultra-compact vision-language model

Controllable TTS with voice cloning, emotion and speed control

Rapid single-image 3D mesh generation

Pure identity face insertion for FLUX

Reranking model

100+ language embeddings

Transformer OCR

Multi-lingual multi-granularity

Multilingual embedding model, 8K context

Task-specific embeddings

Micro model for embedded systems

Ultra-low latency streaming TTS (<97ms)

Efficient DiT architecture

Original PixArt model

Ultra-fast ASR, 60min in 1sec, word timestamps

Segment Anything Model

Conversational TTS with laughter/pauses

Highly expressive TTS

Edge-guided generation

Depth-guided generation

Pose-guided generation

Distilled SDXL - 70% faster

Distilled for speed

Microsoft vision foundation

Balanced ASR model

Fast high-quality ASR

High quality but slower TTS

Neural codec language model TTS

Relighting images with controllable illumination

Fine-tuned turbo whisper for specialized domains

Community favorite, massive ecosystem

Standard inpainting model

Edit images via text instructions

Stable Diffusion for music

Native 768px generation

Native 512px generation

Structure preservation via depth map

Describe voice with text

Mid-tier cascaded model

Single image to 3D in 0.5s

View-consistent 3D generation

Operational AI weather forecasting

AI Global Forecast System

Structured 3D asset generation

Lightweight Apple LLM

Multi-lingual with sound effects

Balanced TTS model

Efficient latent diffusion

Point cloud generation

Single image to 3D views

Lightweight text-only, 32K context

Ultra-light deployment

Compact code completion

Open TTS with pure LLM approach, voice cloning

Scalable 3D generation with structured latents, large variant

47s stereo audio generation (44.1kHz)

1M+ hours multilingual TTS

NVIDIA ASR model

Self-supervised vision

Emotion and prosody control

Text-to-audio generation

Ultra-light edge deployment

Efficient consumer video gen (480p native)

Microsoft's atmospheric foundation model

50% smaller SDXL, 60% faster

OpenAI 3D generation

Tiny reasoning for edge

Tencent DiT foundation model

State-of-the-art embeddings

Instruction-based embeddings

Balanced music generation

Melody-conditioned generation

High-quality audio gen

Ultra-compact coding model

High-quality text embeddings, 8K context

Edge deployment ready

Best open ASR model

Original Moondream

Hyper-realistic dialogue TTS, emotions, voice cloning

On-device assistant specialist

Zero-shot voice design from text descriptions

Few-shot voice cloning with style control