Document Processing

Amazon's highly capable multimodal model balancing accuracy, speed, and cost. Optimized for AWS workloads.

Mistral's 124B multimodal model with frontier-class image understanding. Supports multiple images with 128K context.

Multimodal version of Grok-2 with image understanding capabilities. Processes diagrams, screenshots, and documents.

Hybrid Transformer-Mamba architecture enabling 256K context with efficient processing. Strong multilingual support.

Cohere's most capable model optimized for complex RAG and multi-step tool use. Supports 10 languages.

Balanced model for RAG and tool use at lower cost. Good performance for enterprise applications.

State-of-the-art embedding model for semantic search and RAG. Supports 100+ languages with compression options.