Jamba 1.6 Large is AI21 Labs' flagship hybrid model combining Mamba state-space layers with transformer attention blocks. It supports a 256K token context window and exposes 94 billion active parameters, enabling deep comprehension of long documents and complex multi-document reasoning tasks that overwhelm standard transformer architectures.
AI21 positions Jamba 1.6 Large as a production-grade model optimized for enterprise workloads requiring both long-context fidelity and instruction-following. The SSM-transformer hybrid architecture improves memory efficiency at long contexts compared to pure attention models, making it well-suited to legal, financial, and technical document analysis.
Key Features
256K token context window for long-document and multi-document tasks
Hybrid SSM-Transformer architecture for efficient long-context processing
94B active parameters enabling deep reasoning and nuanced generation
Strong instruction following for structured enterprise workflows
Multilingual text understanding and generation
Function calling and structured output support
Ideal Use Cases
Full-contract legal document analysis and clause extraction
Financial report summarization across lengthy filings
Multi-document research synthesis in knowledge management tools
Enterprise RAG pipelines requiring large context retrieval windows
Long-form technical documentation drafting and review
Example Prompts for Jamba 1.6 Large
Technical Specifications
| Provider | AI21 |
| Category | Text |
| Modality | Text -> Text |
| Context Window | 256K tokens |
Frequently Asked Questions
Try Jamba 1.6 Large now
Start using Jamba 1.6 Large instantly — 100 free credits, no credit card required. Access 750+ AI models through one platform.