AI21 Labs

AI21 Studio: Partially Degraded Service

Status indicator: minor. Partially Degraded Service

Date not specified

AI21 Studio: All Systems Operational

Status indicator: none. All Systems Operational

Date not specified

AI21 Studio Announces Flexible Subscription Model 'Subscribe x whenever'

Date not specified

AI21 Labs adds Markdown export to documentation

AI21 Labs added a dropdown button to documentation pages allowing users to copy content as Markdown or view it as plain-text Markdown. This enhancement improves content reusability across tools and platforms for development teams. No action required - this is a convenience feature addition.

Date not specified

AI21 Maestro integrates with Together AI for open-source model access

AI21 Labs announced a new integration between AI21 Maestro and Together AI, providing native access to Together AI's open-source and frontier models through AI21's platform. This enhancement gives developers greater flexibility, transparency, and cost efficiency when building knowledge agents. No action required - this is a new optional capability.

Date not specified

AI21 Labs releases Jamba Reasoning 3B - open-source reasoning model

AI21 Labs launched Jamba Reasoning 3B, a 3-billion parameter open-source reasoning model with hybrid SSM-Transformer architecture. The model offers 2-5x efficiency gains and can run locally on devices, providing enterprise-ready capabilities for on-device deployment. No action required - this is a new model release expanding available options.

Date not specified

tool_resources parameter deprecated in Maestro runs API

The tool_resources parameter in the Maestro runs API will be deprecated in October 2025 and should be replaced with the tools parameter. Developers need to update their API calls within the next month to avoid potential breaking changes. Migration documentation is available for the new tools parameter.

Date not specified

AI21 Maestro adds Requirements support for File Search and Web Search

AI21 Maestro now allows users to define explicit constraints for format, tone, and content rules when using File Search and Web Search tools. This enhancement provides better control over response structure and consistency across external tool usage. No action required - this is an additive feature that improves existing functionality.

Date not specified

AI21 Labs launches Data Connectors for Enterprise Users

AI21 Labs has released Data Connectors functionality for Enterprise users, enabling direct integration with 11 popular business platforms including Amazon S3, Google Drive, Microsoft Teams, Notion, and others. This new feature expands data accessibility options for enterprise customers using AI21 Studio. No action required for existing users.

Date not specified

AI21 Labs adds Mistral model access via Maestro API and playground

AI21 Labs now provides direct access to mistral-small, mistral-8x7b, and mistral-7b models via their Maestro API and playground interface. Users can access these Mistral models using AI21's API key without needing to provide their own credentials. This expands the available model options for AI21 Labs customers.

Date not specified

Google Drive Integration Added to File Library

New feature allows users to import and sync Google Drive files directly into the platform's File Library with support for common formats and real-time updates. The integration adheres to Google API Services User Data Policy and does not use imported data for AI model training. No action required, optional feature for improved workflow integration.

Date not specified

AI21 Labs launches Maestro - advanced enterprise AI system

AI21 Labs released Maestro, a new enterprise AI system that combines RAG, real-time validation, and dynamic execution planning to automate complex business tasks. The system features validated outputs with constraint enforcement and robust RAG capabilities supporting both semantic file search and web search. This is a new product launch with no breaking changes or required actions for existing users.

Date not specified

AI21 Labs releases Jamba 1.7 with enhanced grounding and self-hosted deployment

AI21 Labs launched Jamba 1.7 featuring better context understanding, faster performance through optimization, and new self-hosted deployment options. The update improves question-answering and instruction-following capabilities while maintaining quality. Both Jamba Large 1.7 and Jamba Mini 1.7 are available via AI21 Studio and Hugging Face.

Date not specified

AI21 Labs releases Jamba 1.6 with 256K context window and improved efficiency

AI21 Labs has released Jamba 1.6 featuring enhanced model quality that outperforms leading open models, extended 256K context window for long document processing, and improved response times. The models are available through AI21 Studio and Hugging Face with options for private deployment. This is a new capability release that expands AI21's model offerings without requiring immediate action from existing users.

Date not specified