AI21 Labs in Amazon Bedrock

Build reliable generative AI applications with foundation models from AI21 Labs
Get started in Amazon Bedrock

Introducing the Jamba 1.5 Family

Per AI21, Jamba 1.5 Large and Jamba 1.5 Mini both focus on speed and efficiency, being 2.5 times faster than many other models due to its hybrid architecture (Transformer and Mamba frameworks).

Jamba 1.5 Large

Jamba 1.5 Large excels at complex reasoning tasks across all prompt lengths, making it ideal for applications that require high quality outputs on both long and short inputs.

Jamba 1.5 Mini

Jamba 1.5 Mini is optimized for low-latency processing of long prompts, enabling fast analysis of lengthy documents and data.

Benefits

Solve core business tasks with purpose-built models that power text generation, lengthy document summarization, and question answering.
Select between Jamba-Instruct, Jurassic-2 Mid, and Jurassic-2 Ultra models based on the complexity of your task, context length needs, and considerations.
Go from prototype to production with expert guidance from AI21 solution architects and account executives.

Meet AI21 Labs

AI21 Labs specializes in developing advanced FMs and AI Systems that enable enterprises to leverage generative AI in production. With their models, AI21 Labs aims to equip businesses to build AI solutions that foster clear communication, confident decision-making, and unrestrained innovation, empowering companies to thrive in the era of artificial intelligence.

Use cases

Summarize the key points from complex and lengthy documents, such as company reports and market analysis, as well as generate term sheets that are already formatted and ready to share with stakeholders.

Generate product descriptions and conversion-optimized marketing content at scale, all written in your brand’s tone, length, and style.

Provide customers with instant, natural language responses to inquiries based on documents, policies, and info sheets.

Boost productivity by making it easy for teams to find synthesized, grounded answers from complex documents or policies using natural language.

Model versions

Jamba 1.5 Large

Jamba 1.5 Large is the ideal model for enterprises tackling data-intensive tasks and excels at complex reasoning for both short and long prompts.

Max tokens: 256K

Languages: English, Spanish, French, Portuguese, Italian, Dutch, German, Arabic and Hebrew

Fine-tuning supported: No

Supported use cases: Instruction following, text generation, document summarization and analysis, question answering

Read the blog

Jamba 1.5 Mini

Jamba 1.5 Mini is a cost-effectove and compact, yet capable model that can tackle a wide range of enterprise-level text generation and comprehension tasks with efficiency. 

Max tokens: 256K

Languages: English, Spanish, French, Portuguese, Italian, Dutch, German, Arabic and Hebrew

Fine-tuning supported: No

Supported use cases: Instruction following, text generation, document summarization and analysis, question answering

Read the blog

Jamba-Instruct

An instruction-tuned version of AI21's hybrid SSM-Transformer Jamba model, Jamba-Instruct is built for reliable commercial use, with best-in-class quality and performance.

Max tokens: 256K

Languages: English

Fine-tuning supported: No

Supported use cases: Instruction following, Text generation, document summarization, question answering

Read the blog

Jurassic-2 Mid

AI21’s mid-sized model for advanced text generation tasks that require both quality and affordability.

Max tokens: 8,192

Languages: English

Fine-tuning supported: No

Supported use cases: Question answering, summarization, draft generation, advanced information extraction, ideation for tasks requiring intricate reasoning and logic.

Read the blog

Jurassic-2 Ultra

AI21’s most powerful model for complex text generation tasks that require the highest quality output.

Max tokens: 8,192

Languages: English, Spanish, French, German, Portuguese, Italian, and Dutch

Fine-tuning supported: No

Supported use cases: Question answering, summarization, draft generation, advanced information extraction, ideation for tasks requiring intricate reasoning and logic.

Read the blog