Hugging Face API
Search and retrieve models, datasets, and AI research papers from the Hugging Face hub — clean JSON, no HF account required, one API key for everything.
What Is Hugging Face Data?
Hugging Face is the central hub for the AI/ML community, hosting over 900,000 models and 200,000 datasets contributed by researchers, enterprises, and independent practitioners. It is also the daily index for AI research papers — every significant paper that lands on arXiv is tracked here with linked models, datasets, and spaces that implement its ideas.
Anysite’s Hugging Face endpoints give you programmatic access to that catalog. Look up a specific model or dataset by its repository id, search the full hub by keyword or tag, or pull the day’s trending AI papers — all via a single POST request with a consistent JSON response. No Hugging Face account, no token setup, no rate-limit management.
All Hugging Face Endpoints
Datasets
| Endpoint | What It Returns |
|---|---|
/api/huggingface/datasets | Dataset details by repository id or URL — id, author, download count, like count, trending score, tags, description, card data, file list |
/api/huggingface/datasets/search | Search datasets by keyword, author, or tag filters — returns ranked list with full metadata including task category, language, and license |
Models
| Endpoint | What It Returns |
|---|---|
/api/huggingface/models | Model details by repository id or URL — id, author, download count, like count, trending score, pipeline tag, library name, tags, config, safetensors info, linked spaces |
/api/huggingface/models/search | Search models by keyword, author, or pipeline/library tags — returns ranked list with architecture details and usage statistics |
Papers
| Endpoint | What It Returns |
|---|---|
/api/huggingface/papers | Paper details by arXiv id or URL — title, authors, summary, AI-generated summary and keywords, upvote count, linked models, datasets, and spaces |
/api/huggingface/papers/search | Search AI research papers by keyword — returns papers with authors, summaries, linked artifacts, and upvote metrics |
/api/huggingface/papers/trending | Daily trending AI research papers — ranked list with linked models, datasets, and spaces; optionally filtered to a specific date |
What You Can Build
AI Model Discovery
Search 900,000+ models by task (text-generation, image-classification, translation), library (transformers, diffusers), or author. Pull trending scores, download counts, and architecture details to evaluate candidates without opening the Hugging Face UI.
Research Paper Monitoring
Pull today’s trending AI papers and their linked models and datasets. Track which papers are gaining upvotes, identify implementations of a specific technique, and surface the week’s most-followed research — all automatically.
Dataset Curation & Audit
Search datasets by task category, language, or license. Retrieve full card metadata — including gated status, storage size, and sibling file lists — to evaluate suitability before ingestion into a fine-tuning or evaluation pipeline.
Competitive ML Intelligence
Monitor a competitor’s Hugging Face organization: new model releases, dataset publications, and linked research. Track trending scores and download counts over time to gauge adoption of specific architectures or checkpoints.
Quick Start
curl -X POST https://api.anysite.io/api/huggingface/models/search -H "access-token: YOUR_API_KEY" -H "Content-Type: application/json" -d '{
"search": "llama",
"filter": ["text-generation"],
"count": 5
}'curl -X POST https://api.anysite.io/api/huggingface/papers/trending -H "access-token: YOUR_API_KEY" -H "Content-Type: application/json" -d '{
"count": 10
}'curl -X POST https://api.anysite.io/api/huggingface/datasets -H "access-token: YOUR_API_KEY" -H "Content-Type: application/json" -d '{
"dataset": "stanfordnlp/imdb"
}'Plans
Every Hugging Face endpoint is included in all Anysite plans — there’s no per-endpoint pricing to track. Start with MCP Unlimited at $30/mo for flat-rate access through your AI tools, or use a credit plan for REST & CLI at scale.
| Plan | Price | Best for |
|---|---|---|
| MCP Unlimited | $30/mo | Unlimited access via Claude, Cursor & ChatGPT (fair use) |
| Starter | $49/mo | REST & CLI — includes a 7-day trial |
| Growth | $200/mo | Production workloads |
| Scale · Pro · Enterprise | from $300/mo | High volume & higher rate limits |
Frequently Asked Questions
You can access model details and search (900,000+ models with metadata, pipeline tags, download counts, and linked spaces), dataset details and search (200,000+ datasets with tags, card data, and file lists), and AI research papers — individual paper lookup by arXiv id, keyword search, and the daily trending papers list with linked models and datasets.
No. Anysite authenticates against the Hugging Face public API using its own credentials. You need only your Anysite API key — no Hugging Face account, no token, and no OAuth setup is required to use any of these 7 endpoints.
Use /api/huggingface/models/search with the filter array parameter. Pass values such as "text-generation", "transformers", or a language tag to narrow results. Combine a search string (e.g. "llama") with filters to find specific model families within a task.
No. These endpoints cover public Hugging Face data only. The response includes a gated field indicating whether a model or dataset requires HF authentication to download its weights, but the metadata itself is always accessible for public gated repositories.
Use /api/huggingface/papers and pass the arXiv id (e.g. "1706.03762" for Attention Is All You Need) or the full arXiv URL. The response includes an AI-generated summary, author list, upvote and comment counts, and arrays of linked models, datasets, and Spaces that implement the paper.
Yes. The MCP Unlimited plan ($30/mo) gives flat-rate access to all Anysite endpoints — including all 7 Hugging Face endpoints — directly inside Claude, Cursor, ChatGPT, and any other MCP-compatible AI tool. Search for models, look up a dataset, or pull today’s trending papers without leaving your AI assistant.
Related Endpoints
Access Hugging Face Data via API or AI Tool
Seven endpoints. One API key. Every model, dataset, and research paper on the Hugging Face hub — via REST, CLI, or directly inside Claude and Cursor.