Popular Model Marketplaces AI tools

16 category leaders in Model Marketplaces, selected from the full directory.

Replicate

  • Hosted open-source models with simple HTTP APIs.
  • Run diffusion, LLM, and vision models without managing GPUs.
  • Versioned model pages and per-second billing.
  • Widely used for fast product prototypes and demos.

Hugging Face

  • Largest open model hub for weights and datasets.
  • Hundreds of thousands of community and org models.
  • Model cards, Spaces demos, and Inference API.
  • Default registry for open-source ML development.

AWS Bedrock

  • AWS managed access to Claude, Llama, and more.
  • Fine-tuning, guardrails, and private endpoints.
  • Enterprise IAM, VPC, and compliance integrations.
  • Default model layer for AWS-native AI apps.

Azure AI Model Catalog

  • Microsoft Foundry catalog with 11,000+ AI models
  • OpenAI, Anthropic, Cohere, Meta, Mistral, DeepSeek, xAI, Black Forest Labs
  • Deploy via PayGo, Managed Compute, or Provisioned Throughput
  • Enterprise model routing and cost optimization

OpenRouter

  • Unified API to hundreds of LLMs from 60+ providers.
  • OpenAI-compatible endpoint; swap models with one key.
  • Pay-as-you-go credits with automatic provider fallbacks.
  • Standard gateway for multi-model app development.

Together

  • Large serverless catalog of open LLMs and multimodal models.
  • Pay-per-token inference plus fine-tuning and GPU clusters.
  • Production path for open-weight model serving at scale.
  • Core infra choice for teams shipping on open models.

Fireworks

  • Fast inference API for open LLMs and image models.
  • Production compound AI systems and LoRA serving.
  • Low-latency serverless with enterprise options.
  • Leading alternative to self-hosting open weights.

Civitai

  • Largest community hub for Stable Diffusion models and LoRAs.
  • Browse, share, and download checkpoints with previews.
  • Creator tools and resources for image model fine-tunes.
  • Default marketplace for open image-generation weights.

NVIDIA NIM

  • NVIDIA catalog of optimized NIM model APIs and containers.
  • Hosted try-and-call endpoints plus self-hosted microservices.
  • Covers LLMs, vision, speech, and retrieval models.
  • Enterprise default on the NVIDIA inference stack.

Google Model Garden

  • Google Cloud catalog of foundation and partner models.
  • Discover, customize, and deploy on Vertex / Gemini.
  • Enterprise peer to Bedrock and Azure model catalogs.
  • Hyperscaler model marketplace for Google Cloud teams.

Fal

  • API catalog for image, video, 3D, and audio models.
  • Serverless endpoints tuned for generative media.
  • Wide diffusion and multimodal model selection.
  • Standard hub for creative-model inference APIs.

Deep Infra

  • Inference APIs for popular open ML models.
  • Serverless LLMs and multimodal serving at scale.
  • Pay-per-use pricing for production workloads.
  • Open-model catalog aimed at app developers.

ModelScope

  • Alibaba open model hub for LLMs and multimodal models.
  • Model cards, fine-tuning, and community downloads.
  • Large catalog spanning Chinese and global open weights.
  • Primary open-model hub alongside Hugging Face.

Featherless

  • Serverless hosting for tens of thousands of open models.
  • One API key across Hugging Face-scale catalogs.
  • Flat-rate plans without self-managed GPUs.
  • Browse-and-run marketplace for open-weight inference.

Ollama

  • Run open-weight LLMs locally on Mac, Windows, and Linux.
  • Simple CLI and API to pull and serve models offline.
  • Default local runtime for developers and hobbyists.
  • Category leader for on-device open-model inference.

Novita

  • Serverless APIs for open and popular generative models.
  • Fast LLM, image, and multimodal inference endpoints.
  • Large model catalog with pay-as-you-go developer access.
  • Major Asia-based alternative to Replicate-class hosts.