Glob.AI OS Version 2026-05 was officially released on July 14, 2026 and includes additional updates delivered in Hot Fix 1 (July 16), Hot Fix 2 (July 17), Hot Fix 3 (July 20) and Hot Fix 4 (July 21).
- API
- Enhanced media parsing reliability for MP4 file uploads related to the omni-parser component.
- Console
- Refined parameter management to provide a smoother experience when accessing the Parameter List.
- Added configurable duplicate validation for parameter insert and update operations.
- Duplicate verification remains enabled by default.
- Organizations can disable the check when needed for performance-sensitive scenarios.
- LLMs
- Added new AWS Bedrock chat models for validation before general availability:
- glm-4.7-flash
- qwen3-next-80b-a3b
- minimax-m2.5
- deepseek-v3.2
- Updated the AWS Bedrock validation catalog to reflect current model availability.
- DeepSeek R1 is now hidden from the validation catalog.
- Agentic Hub
- Enhanced Flows request handling for chat messages containing encoded JSON.
- Improves compatibility with webhook messages that include JSON content in the user message.
- Agentic Hub - Home
- Improved Home to provide a usage-adaptive design with two states:
- Initial state: When there is little interaction with the Agentic Hub, the page displays a selection of featured AI solutions organized into three categories: New Arrivals, Popular, and Top Rated; along with a Trending Agents section in the right-hand sidebar.
- Continue working: As the user interacts with the Agentic Hub, the page evolves to display relevant content based on their recent activity.
- Agentic Hub - Discover
- Improved My Creations to make it easier to understand the publication progress, solution type, and ownership for all solutions in the Project.
- The statuses “Pending Review” and “Pending Changes” will be added to Solution Status.
- You’ll be able to filter by “Agent,” “Flows,” “RAG,” or “Assistant” using Solution Type.
- Filter the different AI solutions by ownership.
- Improved list view to facilitate collaboration.
- Added execution from My Creations for active statuses, including pending review states.
- Optimized conversation creation from My Creations to prevent forbidden errors.
- Improved status display across My Creations and conversations for pending review workflows.
- Agentic Hub - Build
- Optimized the publish workflow to ensure Agents and Flows include a visual identity in Solutions.
- New dedicated Flow avatar generation during the publish flow.
- You are now able to generate an avatar before publishing a Flow, extending the existing capability that was already available for Agents.
- Improved Agent publishing to require an avatar before publication.
- Allows you to generate or replace the avatar directly during the publish steps.
- Refined taxonomy selection to make publishing more flexible.
- Makes taxonomies optional for both Agents and Flows.
- Allows publication without category selection for general-purpose solutions.
- Expanded Agent cards with separate build and publication status tags.
- Refined publish dialogs to surface API validation errors more clearly.
- Strengthened version search to support Agent version names.
- Improved Agent creation so missing model warnings no longer appear before configuration is complete.
- Improved first-time save and publish behavior for newly created Agents.
- Optimized avatar generation and display to preserve the correct image proportions.
- Refined Agent details to preserve line breaks in purpose text.
- Optimized Agent configuration reliability when updating execution permissions.
- Strengthened Agent save reliability when working with nested Agents.
- Enhanced publish validation to block Agents with deprecated, invalid, or missing models.
- Optimized image avatar generation reliability for Agent and flow publication flows.
- Refined Agent name validation to enforce standard supported characters.
- Expanded Agents dashboard loading to avoid empty or partial lists on first access.
- Streamlined Agent chat layout to prevent horizontal misalignment with long links.
- Improved older Agent conversation layouts to prevent horizontal overflow.
- Streamlined Iris loading reliability for Agents, providing a more consistent editing experience.
- Improved public tool metadata with display names for friendlier labels.
- Optimized publish flows with support for extended public descriptions up to 500 characters.
- Expanded trial project initialization to use configurable base Agent definitions.
- Strengthened the trial project experience with a more consistent interface.
- Improved the Integrations and Flows experience with a more consistent interface and behavior.
- New URL-based MCP Server import in Integrations to register remote MCP servers by pasting a server URL. You can onboard external MCP endpoints faster without manual configuration.
- Improving Sharepoint Integration to extract as much text as possible from compatible files, even those that contain only images, by applying appropriate preprocessing to the files.
- Agentic Hub - Run
- Improved conversation streaming to show tool execution status and responses during chat.
- Optimized Agent execution reliability for cases that could return blank responses.
- Improved reliability of file uploads during Agent execution.
- Improved support for scanned and image-based PDFs with compatible models.
- Updated supported model capabilities to include file input where available.
- Improved handling of small file uploads for affected Claude Sonnet models.
- Agentic Hub - Monitor
- Enhanced Flow observability in Session Activity so you can inspect Flow interactions from the same unified view used for Agents
- Improved access to chat transcripts for sessions executed with Flows.
- Improved access to execution logs to help trace and debug Flow behavior.
- Updated the AI solutions listing to show Flows alongside Agents for unified discovery and observability.
- Deprecated legacy Flows analytics in favor of Session Activity.
- Updated the Analytics section to guide users to Session Activity.
- Agentic Hub - Manage
- Improved analytics view in Agentic Hub - Manage - Solutions with summary cards showing metrics for active, pending approval, and unpublished solutions.
- Optimized Admin dashboard to display analytics alongside existing filters and tables.
- Extended publication flows so embedding generation failures no longer block solution publication in incompatible environments.
- Intelligence Hub
- Improved model management to prevent disabling the last active model in a project.
- Streamlined provider management so disabled providers can be re-enabled correctly by authorized users.
- Improved model configuration responses with a flag indicating when only one active model remains.
- New guardrail evaluation to inherits the user ID from the parent request in nested execution scenarios.
- Improved model capability metadata for file-based PDF processing.
- New transport and fallback configuration for gRPC and HTTP/1.1 compatibility.
- LLMs
- New supported models across multiple providers.
- New Anthropic claude-sonnet-5.
- New Vertex AI claude-sonnet-5.
- New Vertex AI claude-fable-5
- New AWS Bedrock claude-sonnet-5.
- New Azure AI Foundry claude-sonnet-5.
- New xAI grok-build-0.1.
- New Vertex AI gemma-4-26b-a4b-it-maas.
- New Vertex AI gemini-embedding-2.
- New AWS Bedrock amazon.titan-embed-image-v1.
- Improved Vertex AI Vector Search support with optional gRPC transport.
- Deprecated legacy Azure OpenAI and Azure AI Foundry models with redirects to newer replacements.
- Deprecated legacy Vertex AI models with redirects to newer replacements.
- Deprecated globant_dgx/GLM-4.6. Migration path: globant_dgx/GLM-5.2.
- Updated globant_dgx/GLM-5.2 routing with transitional fallback support.
- API
- New file metadata lookup by file name to simplify retrieval of uploaded conversation files.
- New organization creation flags to control invitation emails and default Agent creation.
- New endpoint to resolve an internal Agentic ID from an external ID.
- Improved Access Control API to allow project member removal with an organization token.
- Improved Chat API to accept a parent request ID for nested request tracking.
- Optimized Chat API webhook calls to include request identifiers for traceability.
- Strengthened analytics filtering accuracy.
- Expanded runtime policy retrieval for a more reliable experience when using OAuth authentication.
- Streamlined project creation validation to return proper client error responses for invalid requests.
- Improved organization deletion handling to avoid unexpected internal server errors.
- Added pagination support to reasoning strategy listings for easier navigation through large result sets.
- Optimized quota error handling in Chat API to return a specific quota-exceeded response.
- Improved add-user-to-project responses to use consistent userEmail naming.
- Strengthened API token validation to enforce standard supported characters in names and descriptions.
- Updated Data Analyst references to remove disabled assistant types from API responses.
- Console
- Prompt files are now served through authenticated, access-controlled endpoints.
- Improved operation log performance for faster and more reliable data retrieval.
- Optimized Request Exporter access for authorized users.
- Expanded parameter management to prevent duplicate values across scopes.
- Updated Console to hide the RAG Assistant option unless enabled by configuration.