AI Model Catalog

Search and export a current catalog of AI models as an Apify dataset. Each run fetches the latest available metadata, applies your filters, and returns one normalized record per model.

No API key is required. The catalog is refreshed at run time, so pricing, availability, and capabilities should be treated as changeable information and verified with the provider before production use.

What you get

Every dataset item can include:

• Model ID, name, provider, family, and description
• Context window and maximum completion token limits
• Supported input and output modalities
• Token pricing normalized to USD per million tokens
• Tool calling, reasoning, structured output, vision, audio, and video flags
• Supported parameters, model weights, benchmarks, release dates, and expiration dates when available
• A retrieval timestamp and the catalog URL used for that run

The run also saves a RUN_SUMMARY record in the default key-value store with the number of models received, matched, and written, together with the filters used.

Filtering and sorting

Use the input fields to build a focused model shortlist:

• query searches model IDs, names, providers, and descriptions. Example: vision
• providers accepts one or more provider IDs. Example: ["openai", "anthropic"]
• modalities requires input support for every selected modality. Example: ["image"]
• minContextLength keeps models with at least the requested context window. Example: 100000
• maxInputPricePerMillion and maxOutputPricePerMillion set USD-per-million-token limits
• freeOnly keeps models with zero input and output token pricing
• requiresToolCalling, requiresReasoning, and requiresStructuredOutput filter by capabilities
• includeExpired includes models whose catalog expiration date has passed
• sortBy, sortDirection, and limit control ordering and result size

Leave filters empty to return the first 1,000 matching records. Set limit to 0 to write the full available catalog; full exports are written in batches to keep the run efficient.

Common uses

• Compare model pricing and context limits across providers
• Find models suitable for vision, audio, video, or long-context workloads
• Build a budget-conscious shortlist for an application
• Feed normalized model metadata into another workflow, spreadsheet, or internal tool

Output

Results are written to the Actor's default dataset and can be downloaded as JSON, CSV, Excel, or other supported formats. The Output tab includes a readable overview table, while the full records remain available through the dataset API.

Ready-to-use Tasks

Use these public Tasks as starting points:

• Browse the full AI model catalog
• Find long-context OpenAI models
• Find free AI vision models
• Find affordable reasoning models

Data freshness

The Actor retrieves model metadata during every run. Provider names, pricing, capabilities, and model availability can change without notice. Use the data for discovery and comparison, then confirm important details with the relevant provider.