Find your starting point.

Multilingual assistant with switchable thinking and non-thinking modes for everyday work and complex reasoning.

Reasoning-focused checkpoint for multi-step mathematics, programming and general logic tasks.

Instruction-tuned multilingual model for assistant conversations and text generation.

Multimodal assistant that accepts text and images for question answering, summaries and visual analysis.

Text-and-vision assistant with function calling and JSON output for document workflows and conversational agents.

Mixture-of-experts coding model with a long native context and function calling for software agents.

Text-to-image model distilled for image generation in one to four steps.

Guidance-distilled text-to-image model for detailed visual generation and creative experimentation.

Diffusion transformer for image creation with an emphasis on prompt understanding and typography.

Vision-language assistant for interpreting images, documents and video with text responses.

Compact multimodal model for questions about images and videos, visual comparison and captioning.

Multilingual speech-recognition checkpoint for turning recorded audio into text.

Lightweight multilingual text-to-speech model for adding spoken output to applications.

Multilingual retrieval model combining dense, sparse and multi-vector representations for long documents.

Long-context text embeddings with adjustable output dimensions for search and document retrieval.

Compact sentence and short-paragraph encoder producing 384-dimensional vectors for semantic search.