## 1.3.0 (2026-11-03) ### Features * Added new AI models to the default configuration The default configuration now also includes: GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, Claude Opus 5.5. * Added hooks Chat turns can now be delegated to your own agent. With the `turn.start` hook enabled, each turn goes to the configured endpoint before the model runs. The endpoint either answers the turn itself or lets the assistant run with added instructions and context. * Added the AI Diagnostics panel AI Diagnostics allows users to browse the traces and turns of our agent. This helps investigate failures and in-depth problems of LLMs, as well as bugs and other problems. The panel also allows exporting a debuggable data set manually, which can be sent to the Customer Success team to identify areas for improvement. ### Fixes * Fixed a document processing bug that removed content in specific scenarios When a document processing edit unwraps a list, table or blockquote, the content is now properly preserved. * Fixed a document processing bug that misaligned selected parts of the document Document selections in document processing are correctly resolved. * Fixed adding self-closing SVG and MathML elements Self-closing SVG and MathML elements are now properly accepted in edits. * Fixed a bug that removed whitespace from code blocks Whitespace is now properly preserved in code blocks. ### Other changes * Enabled AI spans by default The `LLM_TELEMETRY_ENABLED` option now defaults to `true`. Set it to `false` to stop recording AI requests. * Updated dependencies Selected dependencies have been updated to their latest versions. ## 1.2.1 (2026-10-23) ### Other changes * Updated Node.js to 24.22.1   The Node.js runtime has been updated to 24.22.1, which fixes a high-severity security vulnerability in Node.js core (CVE-2026-111497). * Updated dependencies Selected dependencies have been updated to their latest versions. ## 1.2.0 (2026-09-22) ### Features * Added new AI models to the default configuration The default configuration now also includes: Claude Opus 5, Gemini 3.7 Flash, Gemini 3.8 Flash. * Added more authentication options for Amazon Bedrock and Google Vertex AI Bedrock accepts the standard AWS credential provider chain and Vertex AI accepts Application Default Credentials, so both providers can run on the credentials of the machine or the workload identity instead of static keys. Both also accept assumed-role credentials. * Added an API for managing MCP servers per environment MCP servers scoped to an environment can be created, listed, updated and removed through the API instead of the deployment configuration file alone. The endpoints needed for OAuth-backed servers are available as well. * Added explanations to license errors A license error states its reason in the API response, and the error documentation covers the reasons it can return. ### Fixes * Fixed actions, reviews failing on adaptive Claude models Adaptive-thinking Claude models reject a request that explicitly disables reasoning, which made some actions and reviews fail for specific models. Now the service handles these requests correctly. * Fixed Claude Opus 4.8 reviews on Amazon Bedrock We fixed a bug that caused review calls with Claude Opus 4.8 to fail on Amazon Bedrock provider. * Fixed the quality of edits in specific scenarios We introduced fixes for bugs that caused the model to fail to swap specific elements in document or attach attributes to elements. ### Other changes * Updated dependencies Selected dependencies have been updated to their latest versions. ## 1.1.0 (2026-08-11) ### Features * Added new AI models to the default configuration The default configuration now also includes: Claude Sonnet 5, Claude Fable 5, GPT-5.6 Sol, GPT-5.6 Terra, GPT-5.6 Luna, Gemini 3.6 Flash. * Added Azure OpenAI configuration options Azure OpenAI can now authenticate with tokens instead of API keys. An optional `useDeploymentBasedUrls` flag routes requests through the deployment-based Chat Completions path, which unblocks Foundry-style endpoints that expose a custom base URL. * Added a custom content moderation adapter Deployments can route content moderation to their own service instead of the built-in provider moderation. * Added guardrail rule overrides Each guardrail operation accepts `rules`, which replaces the built-in classification rules, and `extraRules`, which is appended to them. Guardrail checks now also cover context library prompts and take the editor configuration into account. * Added an MCP server for external agents The service exposes its own MCP endpoint, so external agents can process documents, discover and use context items. * Added deployment-level MCP server configuration MCP servers can be configured once for the whole deployment instead of per request, including servers behind OAuth. MCP servers can also back context items, so context content is pulled from external systems such as wikis, document stores and file servers instead of being uploaded. * Added context library Contexts are reusable buckets for prompts and files, defined once in the context library and applied across the different features. A context can hold tone of voice, brand and style guidelines, domain knowledge, or any instruction that would otherwise be repeated in every request. Once saved, a context can be applied to chat messages, actions, and reviews, so the AI follows the same rules consistently without them being restated. * Improved AI editing capabilities Structural editing tools are enabled, including a dedicated table tool, and they can operate on a range of a document instead of its whole content. * Added per-request headers for image analysis Requests can pass `extraHttpHeaders`, which the assistant forwards when fetching images, so images behind authentication can be analyzed. * Added retention cleanup for actions and reviews Stored actions and reviews are now removed by the scheduled cleanup job after one year, the same retention window conversations already use. * Enhanced the document processing endpoint The endpoint can stream its response, with HTML as the default streaming format, process multiple documents and selections in a single request, return changes in the ARF format, accept the editor configuration, use skills, and call external MCP servers. ### Fixes * Fixed partial translation of long documents In specific cases, our service failed to translate some paragraphs in long documents. We have made a number of improvements to our tools to alleviate this problem. * Fixed web resources being returned with web search disabled GPT models no longer return online resources when web search is turned off. * Fixed unsupported file errors being hidden The error is returned in the API response instead of appearing only in the container logs. * Fixed the status code for malformed request bodies A malformed JSON body now returns 400. * Fixed conversation titles for chat commands Titles are generated in the right language and describe the conversation more accurately. * Fixed context identifier validation Identifiers built only from symbols are rejected, and the denylist matches substrings. ### Other changes * Updated dependencies Selected dependencies have been updated to the latest versions, improving performance and security. ## 1.0.0 (2026-06-30) Initial release of the TinyMCE AI On-Premises. ## 0.1.0 (2026-05-26) Early access release of the TinyMCE AI On-Premises.