DeepInfra Models Scraper

What does DeepInfra Models Scraper do?

DeepInfra Models Scraper extracts the complete catalog of ML inference models available on DeepInfra — no API key, no login, and no coding required. Run the actor and get structured data for every model including pricing per million tokens, context window sizes, model type, creator/organization, and model metadata.

The actor fetches DeepInfra's public models page using a single HTTP request and extracts the embedded __NEXT_DATA__ JSON. No browser automation, no Playwright, no proxy required. Every model is returned as a clean JSON record ready to export to CSV, Google Sheets, or any downstream pipeline.

Use this actor to monitor DeepInfra model pricing changes, compare inference costs against other providers (Groq, Fireworks, Together AI, OpenRouter), or automate competitive analysis of fast-inference LLM platforms.

---

Who is it for?

🤖 AI developers and backend engineers

• Find the exact model ID and current pricing before integrating DeepInfra into your application
• Verify context window sizes and supported modalities for planning token budgets
• Automate checks for new model releases or pricing changes

📊 ML researchers and data scientists

• Track DeepInfra pricing trends over time by scheduling recurring runs
• Compare input vs output token prices across text generation, embedding, and multimodal models
• Build datasets for competitive pricing analysis across inference providers

💰 Cost optimization and FinOps teams

• Compare DeepInfra token prices to optimize your inference provider selection
• Monitor price changes across hundreds of hosted models
• Benchmark DeepInfra rates against Together AI, Groq, Fireworks, and other fast-inference providers

🏢 AI product managers and strategists

• Track which new models DeepInfra adds (Llama, DeepSeek, Qwen, Gemma, and more)
• Monitor when models are deprecated and what they are replaced by
• Build dashboards comparing DeepInfra's inference pricing to the broader market

---

Why use DeepInfra Models Scraper?

• No API key required — DeepInfra's models catalog is fully public
• Single HTTP request — fetches all 290+ models in one call with zero JS rendering
• Zero proxy cost — no browser automation or residential proxies needed
• Covers all model types — text generation, embeddings, text-to-image, text-to-speech, automatic speech recognition, text-to-video, reranker, and more
• Structured pricing fields — separate fields for input/output token prices in both cents-per-token and USD-per-million-tokens formats
• Filter by type — narrow results to just the model types you need
• Deprecated model tracking — optionally include deprecated models with replacedBy field to track model lineage
• Pay-per-event pricing — pay only for models extracted, not idle compute time
• Schedule and automate — run daily or weekly to track pricing and model changes over time
• Export anywhere — JSON, CSV, Excel, Google Sheets, or push via API and webhook

---

What data does it extract?

For each model on DeepInfra, the actor extracts:

Field	Description
modelId	Full model identifier (e.g., meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8)
name	Display name (e.g., DeepSeek-V4-Pro)
owner	Creator organization (e.g., deepseek-ai, meta-llama)
modelType	Model category (text-generation, embeddings, text-to-image, etc.)
description	Model description
pricing.type	Pricing structure type (tokens, input_tokens)
pricing.centsPerInputToken	Input price in cents per token
pricing.centsPerOutputToken	Output price in cents per token
pricing.inputPriceUsdPerMillionTokens	Input price in USD per million tokens
pricing.outputPriceUsdPerMillionTokens	Output price in USD per million tokens
pricing.cachedInputRateMultiplier	Discount multiplier for cached inputs
maxTokens	Context window size in tokens
tags	Feature tags (tools, json, reasoning, structured-output, etc.)
quantization	Model quantization (e.g., fp4, fp8)
isPartner	Whether this is a DeepInfra partner model
isDeprecated	Whether the model is deprecated
replacedBy	Model ID of the replacement (for deprecated models)
modelUrl	Direct URL to the model page on DeepInfra
scrapedAt	ISO 8601 timestamp of when the data was scraped

---

How much does it cost to scrape DeepInfra models?

DeepInfra Models Scraper uses pay-per-event pricing:

Event	BRONZE price	FREE price
Run started (one-time)	$0.005	$0.005
Per model extracted	$0.001	$0.00115

Subscribers on higher tiers (SILVER through DIAMOND) pay progressively less per model extracted.

Typical run cost (BRONZE):

• Full catalog (~190 non-deprecated models): ~$0.195 (190 × $0.001 + $0.005 start)
• All models including deprecated (~292): ~$0.297 (292 × $0.001 + $0.005 start)

There is no proxy cost — DeepInfra's models page is publicly accessible.

---

How to use DeepInfra Models Scraper

1. Go to the actor page and click Try for free
2. Configure any optional filters (model type, include deprecated)
3. Click Start and wait a few seconds
4. Download results as JSON, CSV, or Excel — or connect to your pipeline via webhook

No API key or login is required.

---

Input parameters

Parameter	Type	Default	Description
modelType	String	(all)	Filter to a specific model type. Options: text-generation, embeddings, text-to-image, text-to-speech, automatic-speech-recognition, text-to-video, reranker, zero-shot-image-classification
includeDeprecated	Boolean	false	When true, includes deprecated models in the output
maxRequestRetries	Integer	3	Number of retry attempts for failed HTTP requests

---

Output example

{
  "modelId": "deepseek-ai/DeepSeek-V4-Pro",
  "name": "DeepSeek-V4-Pro",
  "owner": "deepseek-ai",
  "modelType": "text-generation",
  "description": "DeepSeek V4 Pro is an MoE model with 1.6T total parameters (49B active) and a 1M-token context window. Built for advanced reasoning, coding, and long-running agent tasks.",
  "pricing": {
    "type": "tokens",
    "centsPerInputToken": 0.000174,
    "centsPerOutputToken": 0.000348,
    "inputPriceUsdPerMillionTokens": 1.74,
    "outputPriceUsdPerMillionTokens": 3.48,
    "cachedInputRateMultiplier": 0.08333333
  },
  "maxTokens": 65536,
  "tags": ["structured-output", "openai", "tools", "json", "reasoning", "featured"],
  "quantization": "fp4",
  "isPartner": false,
  "isDeprecated": false,
  "replacedBy": null,
  "modelUrl": "https://deepinfra.com/deepseek-ai/DeepSeek-V4-Pro",
  "scrapedAt": "2026-04-25T08:40:00.000Z"
}

---

Tips and best practices

💡 Schedule for price monitoring — DeepInfra regularly adds models and updates pricing. Schedule weekly runs and compare datasets to track changes automatically.

💡 Use modelType to narrow results — Filter by text-generation for LLMs, embeddings for vector search models, or automatic-speech-recognition for Whisper-style models.

💡 Check cachedInputRateMultiplier — Many models offer cached context pricing. A multiplier of 0.08 means you pay only 8% of the normal input price for cached tokens.

💡 Monitor deprecated models — Enable includeDeprecated and check replacedBy to track model lineage and plan migrations before models are removed.

💡 Combine with other model scrapers — Pair this actor with Groq Models Scraper, OpenRouter Models Scraper, or Fireworks AI Scraper to build a comprehensive cross-provider pricing dashboard.

💡 Use tags for capability filtering — The tools tag means the model supports function calling, reasoning means chain-of-thought reasoning is enabled, and structured-output means JSON schema output is supported.

---

Integrations

Google Sheets — Export results directly to Google Sheets to build a living pricing dashboard. Schedule daily runs to track changes.

Zapier / Make — Connect the actor's webhook to Zapier or Make to trigger alerts when new models appear or prices change.

Apify API — Fetch results programmatically using the Apify API. Store results in a named dataset for easy comparison across runs.

Database export — Use the JSON export to load model data into PostgreSQL, BigQuery, or any other database for long-term trend analysis.

---

API usage

You can trigger this actor via the Apify API and retrieve results programmatically.

cURL

curl -X POST \
  "https://api.apify.com/v2/acts/automation-lab~deepinfra-models-scraper/runs" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"modelType": "text-generation"}'

Node.js (npm install apify-client)

import { ApifyClient } from 'apify-client';

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

const run = await client.actor('automation-lab/deepinfra-models-scraper').call({
    modelType: 'text-generation',
    includeDeprecated: false,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(`Extracted ${items.length} models`);
console.log(items[0]);

Python (pip install apify-client)

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")

run = client.actor("automation-lab/deepinfra-models-scraper").call(run_input={
    "modelType": "text-generation",
    "includeDeprecated": False,
})

items = client.dataset(run["defaultDatasetId"]).list_items().items
print(f"Extracted {len(items)} models")
print(items[0])

---

Use with Claude AI (MCP)

This actor is compatible with the Apify MCP Server, which lets you run it directly from Claude AI, Cursor, VS Code, or any other MCP-compatible client — no coding required.

Claude Code (terminal)

$claude mcp add --transport http https://mcp.apify.com?tools=automation-lab/deepinfra-models-scraper

Claude Desktop / Cursor / VS Code — add to your MCP config JSON:

{
  "mcpServers": {
    "apify": {
      "type": "http",
      "url": "https://mcp.apify.com?tools=automation-lab/deepinfra-models-scraper",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_API_TOKEN"
      }
    }
  }
}

Example prompts:

• "Run DeepInfra Models Scraper and give me all text-generation models under $1 per million input tokens."
• "Use DeepInfra Models Scraper to find models that support function calling (look for 'tools' in the tags field) and sort by output price."
• "Run DeepInfra Models Scraper and compare embedding model prices — show modelId, owner, and inputPriceUsdPerMillionTokens."

---

Legality and terms of service

DeepInfra's models catalog is publicly accessible without a login. This actor only scrapes data that is freely available on deepinfra.com/models. Always review DeepInfra's Terms of Service before using scraped data commercially.

---

FAQ

Does it require a DeepInfra API key? No. The models page is publicly accessible without authentication.

How many models does it return? DeepInfra currently lists 292 models total, ~190 of which are non-deprecated. The count grows as DeepInfra adds new models.

Can I filter by model type? Yes. Use the modelType input parameter to restrict results to a specific category such as text-generation or embeddings.

How often should I run it? Weekly runs are sufficient for most use cases. For production pricing monitoring, daily runs are recommended.

Does it use proxies? No. DeepInfra's public page is accessible without proxies, keeping costs minimal.

---

Related scrapers

• Groq Models Scraper — Groq inference models with speed benchmarks and rate limits
• OpenRouter Models Scraper — 300+ models from OpenRouter with unified pricing
• Fireworks AI Scraper — Fireworks AI inference platform model catalog
• Artificial Analysis Scraper — Cross-provider AI benchmark and pricing comparisons