Ollama Library Scraper

Scrape the entire Ollama model library and extract structured data about every available LLM. Get model names, pull counts, parameter sizes, capabilities, descriptions, and update dates — all in one run.

What does Ollama Library Scraper do?

Ollama Library Scraper is an Apify actor that crawls the Ollama model registry at ollama.com and extracts metadata for every model listed. It handles pagination automatically, letting you scrape from a handful of top models to the entire library in a single run.

The scraper works with plain HTTP requests — no browser needed — making it fast and cost-effective. It parses server-rendered HTML to extract all visible model information directly from the search results page.

Who is Ollama Library Scraper for?

• AI/ML engineers building local LLM pipelines who need to monitor available models and their sizes
• DevOps teams evaluating which models to deploy on-premises
• Researchers tracking the open-source LLM ecosystem growth, comparing model families, and analyzing trends
• Product managers benchmarking the landscape of available models for competitive analysis
• Data analysts building dashboards to track model popularity, new releases, and ecosystem metrics

Why use Ollama Library Scraper?

• No API exists — Ollama has no public API for browsing their model library. This scraper fills that gap.
• Zero competitors — there are no other Ollama scrapers on the Apify Store.
• Fast and cheap — HTTP-only scraping means sub-minute runs for the entire library.
• Structured output — get clean JSON with numeric pull counts, capability tags, and ISO dates.
• Filter and search — narrow results by keyword, capability (vision, tools, embedding, etc.), or sort order.

What data can you extract from Ollama?

Field	Type	Example
name	string	"qwen3"
slug	string	"qwen3"
description	string	"Qwen3 is the latest generation..."
pullCount	string	"27.3M"
pullCountNumeric	number	27300000
tagCount	number	15
parameterSizes	string[]	["0.6b", "4b", "8b", "14b", "30b"]
capabilities	string[]	["vision", "tools", "thinking"]
updatedAt	string	"3 days ago"
updatedAtPrecise	string	"Apr 15, 2026 10:02 PM UTC"
modelUrl	string	"https://ollama.com/library/qwen3"
scrapedAt	string	"2026-04-20T08:17:33.077Z"

How much does it cost to scrape Ollama models?

Ollama Library Scraper uses pay-per-event pricing. You pay only for what you use:

Event	Price
Run started	$0.005 per run
Model extracted	$0.0005 per model

Example costs:

• Scrape top 50 models: $0.005 + 50 x $0.0005 = $0.03
• Scrape all ~500 models: $0.005 + 500 x $0.0005 = $0.26

On the Apify Free plan ($5/month), you can scrape the full library ~19 times per month.

How to scrape the Ollama model library

1. Go to Ollama Library Scraper on Apify.
2. Click Start to scrape all popular models with default settings.
3. Optionally enter a Search keyword (e.g., "llama", "vision", "embedding") to filter models.
4. Set Max models to limit how many results you want.
5. Choose Sort by — Popular (most pulled) or Newest (recently added).
6. Use Filter by capabilities to only get models with specific features (vision, tools, thinking, etc.).
7. Click Start and wait for the run to finish.
8. Download your data as JSON, CSV, Excel, or connect it via API.

Input parameters

Parameter	Type	Description	Default
searchQuery	string	Keyword to search for (leave empty for all models)	""
maxModels	integer	Maximum models to extract (0 = unlimited)	100
maxPages	integer	Maximum search pages to scrape (~20 models/page)	50
sortBy	enum	Sort order: popular or newest	popular
filterCapabilities	string[]	Filter by capabilities: cloud, embedding, vision, tools, thinking, code	[]
maxRequestRetries	integer	Retry attempts for failed requests	3

Output example

{
    "name": "qwen3",
    "slug": "qwen3",
    "description": "Qwen3 is the latest generation of large language models in Qwen series...",
    "pullCount": "27.3M",
    "pullCountNumeric": 27300000,
    "tagCount": 15,
    "parameterSizes": ["0.6b", "1.7b", "4b", "8b", "14b", "30b"],
    "capabilities": ["tools", "thinking"],
    "updatedAt": "6 months ago",
    "updatedAtPrecise": "Oct 15, 2025 3:22 PM UTC",
    "modelUrl": "https://ollama.com/library/qwen3",
    "scrapedAt": "2026-04-20T08:17:33.077Z"
}

Tips for best results

• Start with defaults — the prefilled input scrapes the top 50 most popular models, which covers the most commonly used LLMs.
• Use capability filters to narrow results — e.g., filter by "vision" to find all multimodal models, or "embedding" for vector embedding models.
• Set maxModels to 0 to scrape the complete library without limits.
• Sort by newest to discover recently added models that may not have high pull counts yet.
• Combine search + filters — search for "llama" with "tools" capability to find tool-calling Llama variants.

Integrations with other tools

• Google Sheets — export model data to a spreadsheet for tracking trends over time. Schedule daily runs to build a historical dataset of model popularity.
• Slack notifications — get alerts when new models appear in the library by comparing runs.
• Airtable — build a searchable model catalog with filtering by capability and size.
• Zapier / Make — trigger workflows when specific models are updated or new ones are added.
• Custom dashboards — feed the structured JSON into Grafana, Metabase, or any BI tool to visualize the Ollama ecosystem.

How to use Ollama Library Scraper via API

Node.js

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_API_TOKEN' });

const run = await client.actor('automation-lab/ollama-library-scraper').call({
    searchQuery: '',
    maxModels: 100,
    sortBy: 'popular',
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);

Python

from apify_client import ApifyClient

client = ApifyClient('YOUR_API_TOKEN')

run = client.actor('automation-lab/ollama-library-scraper').call(run_input={
    'searchQuery': '',
    'maxModels': 100,
    'sortBy': 'popular',
})

items = client.dataset(run['defaultDatasetId']).list_items().items
print(items)

cURL

curl "https://api.apify.com/v2/acts/automation-lab~ollama-library-scraper/runs" \
  -X POST \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -d '{
    "searchQuery": "",
    "maxModels": 100,
    "sortBy": "popular"
  }'

Use with MCP (Model Context Protocol)

Connect Ollama Library Scraper to AI assistants using MCP for on-demand model discovery.

Claude Code

$claude mcp add --transport http apify "https://mcp.apify.com?tools=automation-lab/ollama-library-scraper"

Claude Desktop / Cursor / VS Code

Add to your MCP config file:

{
    "mcpServers": {
        "apify": {
            "url": "https://mcp.apify.com?tools=automation-lab/ollama-library-scraper"
        }
    }
}

Example prompts:

• "Show me all Ollama models with vision capability"
• "What are the top 10 most popular models on Ollama?"
• "List all embedding models available on Ollama with their sizes"
• "Find Ollama models updated in the last week"

Is it legal to scrape Ollama?

Ollama's model library is publicly accessible without authentication. This scraper accesses only publicly visible model metadata — the same information any visitor sees when browsing ollama.com/search. The scraper does not download any model weights, bypass access controls, or access private data.

Always review Ollama's terms of service and robots.txt before large-scale scraping. Use reasonable request rates and respect the site's resources.

FAQ

How many models are in the Ollama library? The Ollama library currently contains 500+ models and grows regularly as new open-source LLMs are released. The scraper handles pagination automatically to capture all of them.

Can I filter models by parameter size? The scraper extracts all parameter sizes for each model (e.g., 0.6b, 4b, 8b, 14b). While filtering by size isn't a direct input option, you can filter the output dataset by the parameterSizes field after scraping.

How often is the data updated? Each model listing includes an updatedAt field showing when it was last modified on Ollama. Schedule recurring runs to track changes over time.

What if a request fails? The scraper automatically retries failed requests up to 3 times (configurable) with exponential backoff. If a page consistently fails, it's skipped and scraping continues with the next page.

Why are some pull counts shown as "27.3M" instead of exact numbers? Ollama displays abbreviated pull counts on their site. The scraper provides both the raw display value (pullCount: "27.3M") and a parsed numeric value (pullCountNumeric: 27300000) for sorting and analysis.

Related scrapers

• ArXiv Scraper — scrape research papers from ArXiv
• AI Tools Directory Scraper — scrape AI tool listings