Palmyra X5 is Writer’s most advanced enterprise large language model, designed to power the next generation of AI agents and applications. It features a 1 million-token context window, enabling the model to process and reason over vast amounts of data, such as entire codebases or large document collections, in a single prompt. Palmyra X5 is built with adaptive reasoning, hybrid attention mechanisms, and supports multi-step, agentic workflows. The model is trained entirely on synthetic data, achieving high efficiency and cost-effectiveness, and is available via Amazon Bedrock for seamless enterprise integration. Palmyra X5 supports multiple languages, tool-calling, built-in retrieval-augmented generation (RAG), code generation, structured outputs, and multi-modality, making it suitable for complex, mission-critical enterprise use cases.

Key Features

1 Million-Token Context Window Processes and reasons over extremely large inputs.

• Ingests entire codebases or large document sets
• Enables long-context, multi-step workflows

Adaptive Reasoning and Hybrid Attention Delivers advanced, dynamic reasoning capabilities.

• Supports agentic AI and multi-step task execution
• Maintains accuracy across long contexts

Multi-Modality and Multi-Language Support Handles diverse data types and languages.

• Text, code, and rich text formatting
• English, Spanish, French, German, Chinese, and more

Tool-Calling and LLM Delegation Automates complex workflows and integrates with external tools.

• Enables agentic systems and workflow automation
• Supports LLM-to-LLM delegation for distributed reasoning

Built-in Retrieval-Augmented Generation (RAG) Enhances accuracy with real-time data integration.

• Connects to enterprise data sources
• Improves factuality and relevance

Enterprise-Grade Security and Deployment Designed for secure, scalable enterprise use.

• Available via Amazon Bedrock (serverless deployment)
• Compliance with industry standards

Benefits

Scalable AI for Complex Workflows Handles large-scale, multi-step enterprise tasks.

• Powers agentic AI systems and automation
• Reduces need for prompt engineering and context management

Cost and Efficiency Leadership Optimized for enterprise budgets and performance.

• Trained with synthetic data for $1M in GPU costs
• Among the fastest and most cost-effective large-context LLMs

Enhanced Accuracy and Reliability Maintains high performance on long-context and multi-modal tasks.

• Reduces hallucinations and context loss
• Consistent outputs for mission-critical applications

Seamless Enterprise Integration Easy to deploy and manage in enterprise environments.

• Available through Amazon Bedrock API
• Supports serverless, scalable deployment