Glob.AI OS Version 2026-06 was officially released on August 18, 2026 and includes additional updates delivered in Hot Fix 1 (September 16).
- LLMs
- New OpenAI GPT-5.6 models to expand the available model catalog.
- New Gemini 3.7 Flash support in Google Vertex AI for low-latency generation workloads.
- New Grok 4.6 as a flagship xAI model in the catalog.
- New Qwen 3.8-27B availability through Globant DGX.
- Improved GLM-4.6 availability.
- Updated GPT-5.6 pricing to reflect the latest OpenAI price reduction.
- Restored access to older xAI models.
- Refined token consumption reporting for GLM-4.6 so you can view usage correctly.
- Improved streaming stability for Anthropic models, so responses that use tools are delivered reliably
- Improved Gemini Flash context caching during Agent runs with short prompts.
- Integrations
- Refined MCP Tool name handling to preserve the expected naming format and prevent Tool not found errors in Atlassian Rovo Integrations.
- Agentic Hub - Home
- Appearance & Branding: Adds a new organization-level configuration that lets Organization Administrators customize the appearance of Hubs. You can configure themes and colors for Light and Dark modes, upload logos and a favicon, set the platform name and accessibility text, and configure the default language and languages available to users.
- Added a consistent page-level loading indicator across screens that fetch data on first view, so you get clear feedback instead of a blank screen while content loads.
- Optimized scrolling behavior so the sidebar and top bar stay fixed and only the content area scrolls, consistently across the platform.
- Added Spanish (Spain) as a supported language and renamed the existing Spanish option to Spanish (Latin America).
- Agentic Hub - Discover
- New Navigate Agent available out of the box as a default installed solution, without requiring manual subscription.
- Refined listing pages to reset the filter menu, search text, and sort selection when you change organization or project, across Discover and Manage.
- Refined button styling to apply a consistent oval/pill shape across Discover, Run, and Manage.
- Refined the Solution Detail screen with a sticky header, keeping the solution actions (Run, Options, Favorite) accessible while you scroll.
- Added the ability for editors and read-only members to leave a shared collection, and a "Back to Collections" navigation control on the Collection Detail screen.
- Optimized Collection cards to show the solution type tag (Agent or Flow).
- New Visibility & Execution Scope step users see when publishing to Projects.
- Fixed the Publish Agent dialog rendering with a dark background while in Light mode
- Agentic Hub - Build
- Streamlined Agent and Flow publishing by making the avatar optional.
- You can now publish an Agent or a Flow without generating or uploading an avatar image.
- Corrected the misleading helper text that suggested a random image would be generated.
- Simplified the avatar picker so selecting a generated thumbnail sets it directly, without a separate confirmation step.
- Enhanced Sub-Agent execution to automatically propagate the OAuth token used to invoke the parent Agent to its Sub-Agents' internal Tools, without Per-Integration configuration.
- Added the Conversation ID as an agent context variable, so builders can reference it in tools and prompts.
- Enhanced Agent avatar generation to work reliably across more environments and image providers.
- Enhanced Iris to work with AWS Bedrock models.
- Added a warning when an Agent's Tools are unconfigured, so you can spot and fix them before running or publishing.
- Refined the Tool selection modal so resizing the tools panel to its minimum width no longer causes text overlap or overflow.
- Improved credential management error handling, mapping known errors to specific messages and showing a generic message for unexpected failures.
- Made credential status indicators clearer and more consistent when you enable, disable, or edit a credential.
- Optimized editing of credentials that belong to a deleted integration so the edit view no longer breaks.
- Refined new Flows to use an Agent node in the default template instead of a deprecated Assistant node.
- Optimized spreadsheet name resolution in Flows.
- Added tab persistence for Reasoning Strategies, so your previously open tabs are restored the next time you return.
- Optimized SharePoint Integration to correctly read PDF files, matching the behavior of directly uploaded PDFs.
- Optimized Integration Tool publishing to return a meaningful result when a re-imported integration has no tools to publish.
- Agentic Hub - Run
- You can interact with a Chat with Data Assistant in Run to query and analyze data using natural language, refine your queries, and explore results through interactive visualizations. For more information, see Chat with Data Assistant queries in Run.
- Fixed an issue that prevented downloading generated artifacts from the Files section.
- Optimized the Workspace so messages no longer disappear intermittently during a conversation.
- Optimized the Workspace to run and display the correct, current Agent version.
- Optimized conversation deletion so it succeeds even when artifacts are linked to the conversation.
- Optimized links to RAG source files in conversations so they open correctly.
- Optimized Agentic Gateway reliability by routing to an available, supported model.
- Optimized Agentic Process reliability so a task's output variable is reliably set before downstream conditions run.
- Agentic Hub - Monitor
- Processes in Monitor provides a dedicated view to monitor the execution of Agentic Processes.
- Features Processes list with search, filters, and date-range filtering, with real-time status updates.
- Shows Process detail with Overview and History tabs, including artifacts, activities, and step-level history.
- Stop a running process directly from the list, transitioning it to Cancelled.
- Added empty-state messages for the Processes list, Artifacts, and Activities tables.
- Remembers your open Process tabs and restores them the next time you return.
- Enhanced Session Activity to show failed Flow executions, so you can identify and inspect Flow failures.
- Added tab persistence to Session Activity, restoring your previously open tabs.
- Agentic Hub - Manage
- New deprecated-model management workflow in Moderate Solutions.
- New "Deprecated Model" filter to narrow the solutions list to those using a deprecated model.
- You can now request changes on a published solution when its model is deprecated, moving it back to Pending Changes so the author can update it.
- New Appearance & Branding customization for organizations, available to organization administrators from Settings.
- Theme customization with presets, an advanced per-token color panel for light and dark mode, and a live preview.
- Branding customization for logos, favicon, display name, and per-language alt text.
- Language configuration for the default and supported platform languages.
- Optimized the "Unpublished solutions" metric to also count solutions in a deleted state.
- Refined the Solution Moderation detail with a sticky "Back to admin" bar, and renamed its primary action from "Continue" to "Save".
- Optimized sticky headers on the Taxonomies and Manage Solutions lists so only the list scrolls.
- Intelligence Hub
- New organization-level model overrides to enable or disable models per organization.
- LLMs
- New supported models across multiple providers.
- New Anthropic claude-opus-5.
- New AWS Bedrock claude-opus-5.
- New Vertex AI claude-opus-5.
- New Azure AI Foundry claude-opus-5.
- New Vertex AI gemini-3.6-flash.
- New Vertex AI gemini-3.5-flash-lite.
- New xAI grok-4.5.
- New Cohere rerank-v4.0-fast.
- New Cohere rerank-v4.0-pro.
- New OpenRouter moonshotai/kimi-k3.
- Updated the Azure AI Foundry claude-opus-4-8 context window to 1,000,000 tokens.
- PyGEAI v0.8.0
- The pygeai.proxy module will be deprecated and moved to a standalone package. You must migrate to the new pygeai-proxy package.The proxy functionality for managing MCP and A2A servers has been extracted from the main pygeai package into a dedicated pygeai-proxy package. This change provides better separation of concerns and allows the proxy functionality to evolve independently.
- API
- New parameter management endpoints to Create, Upsert, Get, and Delete parameters, restricted to GAM Administrators.
- New SDK support for programmatically managing API keys.
- Standardized authentication responses so requests without valid credentials return a clear 401 Unauthorized, across the Tools, Runtime, and Job Queue APIs.
- Optimized request attribution so requests sent with a GAM token and project ID preserve that context instead of being associated with an API token.
- Optimized file references in agent Guidelines ({file:...}) so they resolve correctly instead of returning an error.
- Optimized agent execution reliability for cases that could fail due to missing tool call identifiers.
- Optimized token usage tracking to record total token counts even when the associated cost is zero.
- Optimized zero-cost models so they no longer trigger a "Quota Exceeded" error, and their usage is tracked in the dashboard.
- Optimized billing accuracy for cache-write tokens on streaming requests.
- Optimized loading of secret parameter values defined at the organization scope.
- Refined the GET credentials response to use consistent key casing.
- Console
- Optimized the display of the dollar budget option for organizations and projects where it was not appearing.
- Optimized Console navigation so sidebar options respond without long delays when the Chat with Data service is unavailable, and added a clear message when it is off.
- Enhancing Quota Management capabilities.
- Simplifying quota administration and automation for easier usage limit management across the platform.
- Improving aspects to ensure that organizational limits appear correctly in the interface.
- Ensuring that previously configured organizational limits are consistently displayed.
- Preventing discrepancies in limit visibility that could make it difficult for users to monitor and manage resource usage.
- Ensuring that appropriate limit information is displayed.
- Optimized the Secrets page so the full list loads reliably with working pagination.
- Optimized organization creation so name validation no longer returns an internal server error.
- Improved handling of multiple and low-quality or handwritten files in the Chat Assistant.