Model collections

The catalogue by what models do: free models, million-token context, vision, documents, audio, image and video generation, speech, embeddings and rerankers.

The catalogue grouped by what each model accepts and produces, as the catalogue records it. Each collection opens the model list with its filter applied.

• Free models: Models whose every price is zero.
• A million tokens of context: Models that take at least 1,000,000 tokens in one request.
• Vision: Models that take images as input.
• Documents: Models that take files such as PDFs as input.
• Audio input: Models that take audio as input, including transcription.
• Embeddings: Models that turn text into vectors, for search and retrieval.
• Rerankers: Models that score documents against a query.