- **Model provider**: Allows users to modify the model type of existing model configurations. [#16029](https://github.com/infiniflow/ragflow/pull/16029)
- **Chat channel**: Enables users to deploy RAGFlow assistants as chatbots within external messaging platforms like Discord and Feishu. [#15850](https://github.com/infiniflow/ragflow/pull/15850)
- **Observability**: Groups multi-turn chat traces by session within Langfuse to facilitate conversation debugging and replay. [#15679](https://github.com/infiniflow/ragflow/pull/15679)
### i18n
- Adds ~70 missing translation keys for the French UI. [#15983](https://github.com/infiniflow/ragflow/pull/15983)
### Bug fixes
- **Model provider**: Fixes inaccurate token billing, corrects embedding truncation limits to prevent indexing failures, and refactors error handling to surface root error causes. [#15424](https://github.com/infiniflow/ragflow/pull/15424)
- **API**: Adds a legacy mode to the `/chat/completions` endpoint to restore the v0.23.0 style `<think>` output for backward compatibility. [#16014](https://github.com/infiniflow/ragflow/pull/16014) See also [Converse with chat assistant](./references/http_api_reference.md#converse-with-chat-assistant).
- Implements auto-populated model lists for multiple providers, eliminating the need to type model names manually. This feature currently supports: Ollama, OpenRouter, vLLM, OpenAI-API-Compatible, LM-Studio, VolcEngine, Xinference, LocalAI, BaiduYiyan, GPUStack, and Fish Audio.
- Allows configuring multiple API keys for the same model provider. [#14595](https://github.com/infiniflow/ragflow/pull/14595)
- Dynamically populates model selection dropdowns in the UI by fetching the currently available models directly from remote model providers. [#15711](https://github.com/infiniflow/ragflow/pull/15711)
- **Data source connectors**: Implements new data source connectors for Outlook, OneDrive, Microsoft Teams, Slack, SharePoint, Salesforce, and Azure Blob Storage. [#15333](https://github.com/infiniflow/ragflow/pull/15333)[#15330](https://github.com/infiniflow/ragflow/pull/15330)[#15332](https://github.com/infiniflow/ragflow/pull/15332)[#15188](https://github.com/infiniflow/ragflow/pull/15188)[#15190](https://github.com/infiniflow/ragflow/pull/15190)[#15462](https://github.com/infiniflow/ragflow/pull/15462)[#15466](https://github.com/infiniflow/ragflow/pull/15466)
- **Dataset** - Implements a checkpoint and resume feature for community extraction and entity resolution, the most expensive and time-consuming parts of the GraphRAG indexing pipeline. [#15518](https://github.com/infiniflow/ragflow/issues/15518)[#15523](https://github.com/infiniflow/ragflow/pull/15523)
### Improvements
- Removes `<think>` text buffering to ensure reasoning-capable models feel faster and more transparent during interactions. [#15891](https://github.com/infiniflow/ragflow/pull/15891)
- Marks MySQL migrations as applied. [#15504](https://github.com/infiniflow/ragflow/pull/15504)
### Model Support
- Four new SiliconFlow models [#15383](https://github.com/infiniflow/ragflow/pull/15383)
- MiniMax-M3 model [#15513](https://github.com/infiniflow/ragflow/pull/15513)
- Voyage 4 model family [#15516](https://github.com/infiniflow/ragflow/pull/15516)
- Cohere model list. [#15576](https://github.com/infiniflow/ragflow/pull/15576)
### i18n
- Completes Korean translation. [#15863](https://github.com/infiniflow/ragflow/pull/15863)
- Completes Italian translation. [#15729](https://github.com/infiniflow/ragflow/pull/15729)
### Bug fixes
- The thinking mode of MiniMax models was not correctly enabled. [#15496](https://github.com/infiniflow/ragflow/pull/15496)
- Infinite loops were triggered when the thinking mode was enabled for Qwen3.5 and Qwen3.6 models. [#15101](https://github.com/infiniflow/ragflow/pull/15101)
- Streamed answers were being duplicated when using the OpenAI-compatible chat completions API endpoint. [#15286](https://github.com/infiniflow/ragflow/issues/15286)[#15443](https://github.com/infiniflow/ragflow/pull/15443)
- Serialization errors were caused during chat completions when invalid numeric scores like `NaN` (Not-a-Number) or `Inf` (Infinity) were passed to the JSON encoder. [#15245](https://github.com/infiniflow/ragflow/issues/15245)[#15266](https://github.com/infiniflow/ragflow/pull/15266)
- Chat completions using LiteLLM providers were failing because unrecognized internal configuration parameters were not being filtered out before reaching the external APIs. [#15427](https://github.com/infiniflow/ragflow/issues/15427)[#15432](https://github.com/infiniflow/ragflow/pull/15432)
- The OpenAI-compatible chat completions API was defaulting to streamed responses. [#15356](https://github.com/infiniflow/ragflow/issues/15356)[#15394](https://github.com/infiniflow/ragflow/pull/15394)
- Empty `AND` results were incorrectly dropped during metadata filtering. [#15477](https://github.com/infiniflow/ragflow/pull/15477)
- Repetitive page chrome, such as headers and footers, was incorrectly extracted as main text by the MinerU parser. [#15335](https://github.com/infiniflow/ragflow/issues/15335)[#15387](https://github.com/infiniflow/ragflow/pull/15387)
- English chart titles were missing during document extraction in the DeepDoc module. [#15481](https://github.com/infiniflow/ragflow/pull/15481)
- Empty outputs were returned by the TitleChunker for `json` and `chunks` upstream formats [#14247](https://github.com/infiniflow/ragflow/pull/14247)[#15396](https://github.com/infiniflow/ragflow/pull/15396)
- An error message was missing when a .tsv file upload attempt failed. [#15284](https://github.com/infiniflow/ragflow/pull/15284)
- Caps the maximum page size to fix system crashes or slowdowns from large queries. [#15292](https://github.com/infiniflow/ragflow/pull/15292)
- Client errors were caused by the OpenAI-compatible chat completion API incorrectly defaulting to streamed responses. [#15356](https://github.com/infiniflow/ragflow/issues/15356)[#15394](https://github.com/infiniflow/ragflow/pull/15394)
- HTTP 500 internal server errors were triggered instead of standard 4xx client errors when users attempted to download missing files from the storage backend. [#15369](https://github.com/infiniflow/ragflow/issues/15369)[#15371](https://github.com/infiniflow/ragflow/pull/15371)
- GraphRAG entity ranking was broken. [#15795](https://github.com/infiniflow/ragflow/issues/15795)[#15797](https://github.com/infiniflow/ragflow/pull/15797)
- Agent: Adds a **Browser** component that enables AI to autonomously navigate and interact with web pages. [#14888](https://github.com/infiniflow/ragflow/pull/14888)
- RAG: Stabilizes RAPTOR's AHC mode (Ψ-RAG), which was introduced in v0.25.3 to resolve previous semantic loss by building individual document trees and then merging them hierarchically. This new approach significantly accelerates index construction and outperforms the legacy GMM mode in Recall@5 and average F1. Users retain the option to switch between the two modes. [#14674](https://github.com/infiniflow/ragflow/issues/14674)[#14679](https://github.com/infiniflow/ragflow/pull/14679)
- Agent: Introduces lightweight `@tool` decorator to streamline Python function registration process for chat models. [#15047](https://github.com/infiniflow/ragflow/pull/15047)
- Agent: Enables agent messages to display base64-encoded images. [#15212](https://github.com/infiniflow/ragflow/pull/15212)
- Agent: Allows developers to pass `chat_template_kwargs` to agent chat completion endpoint. [#14182](https://github.com/infiniflow/ragflow/issues/14182)[#14542](https://github.com/infiniflow/ragflow/pull/14542) See also [Converse with agent](./references/http_api_reference.md#converse-with-agent)
- Fixes `/chat/completions` to allow sending only latest message in API payload and removes requirement to transmit full conversation history. [#15197](https://github.com/infiniflow/ragflow/pull/15197) See also [Converse with chat assistant](./references/http_api_reference.md#converse-with-chat-assistant).
- Weight assigned to vector similarity was not properly applied during the retrieval phase. [#15108](https://github.com/infiniflow/ragflow/pull/15108)
- Users were unable to save parser configurations on the dataset configuration page. [#15175](https://github.com/infiniflow/ragflow/issues/15175)[#15177](https://github.com/infiniflow/ragflow/pull/15177)
- Log text on a data source's details page was truncated. [#15056](https://github.com/infiniflow/ragflow/pull/15056)
- An unresponsive "Status" filter on the document list page prevented users from filtering or managing uploaded documents by their parsing status. [#15170](https://github.com/infiniflow/ragflow/issues/15170)[#15216](https://github.com/infiniflow/ragflow/pull/15216)
- Calling `GET /agents/<agent_id>/sessions/<session_id>` with a missing or invalid session ID caused a server error. [#14989](https://github.com/infiniflow/ragflow/issues/14989)[#15011](https://github.com/infiniflow/ragflow/pull/15011)
- RAG: RAPTOR construction process halted when using the [Infinity](https://github.com/infiniflow/infinity) document engine. [#14998](https://github.com/infiniflow/ragflow/pull/14998)
- The system failed to correctly parse structured content returned by Mistral reasoning models. [#14805](https://github.com/infiniflow/ragflow/pull/14805)
- The **Parser** component in an ingestion pipeline incorrectly retained raw HTML tags in its text output. [#14831](https://github.com/infiniflow/ragflow/issues/14831)[#14920](https://github.com/infiniflow/ragflow/pull/14920)
- The table parser incorrectly extracted or attached metadata during document processing. [#15127](https://github.com/infiniflow/ragflow/pull/15127)
- Asynchronous background tasks and nested event loops were not properly handled, causing backend instability. [#14755](https://github.com/infiniflow/ragflow/issues/14755)[#14761](https://github.com/infiniflow/ragflow/pull/14761)
- Prompt variables configured in the **Agent** component disappeared after being entered. [#15218](https://github.com/infiniflow/ragflow/pull/15218)
### i18n
- Fully translates the interface into French with the addition of roughly 1,400 localization keys. [#15192](https://github.com/infiniflow/ragflow/pull/15192)
- Introduces local and SSH provider options for sandbox environment settings within the admin interface, allowing administrators to configure execution environments without editing environment variables. [#15039](https://github.com/infiniflow/ragflow/pull/15039)
- Elasticsearch: Accelerates the retrieval process by removing unnecessary vector fetches during the main search phase, reducing latency by 50–100%. [#14970](https://github.com/infiniflow/ragflow/pull/14970)
- Pushes metadata filters down to the [Infinity](https://github.com/infiniflow/infinity) document engine, significantly improving retrieval performance. [#14974](https://github.com/infiniflow/ragflow/pull/14974)
- Introduces Redis-based caching for Text-to-Speech model outputs, eliminating redundant API calls for identical text to reduce latency and save provider quota. [#14851](https://github.com/infiniflow/ragflow/pull/14851)
- Reduces server startup time by 5-9 seconds and saves roughly 200MB of memory by replacing heavy module-level imports with lazy runtime loading. [#14973](https://github.com/infiniflow/ragflow/pull/14973)
- Optimizes the connector dashboard. [#14979](https://github.com/infiniflow/ragflow/pull/14979)
- Atomic database updates: Wraps document and dataset chunk counter updates in atomic database transactions to prevent data drift. [#14866](https://github.com/infiniflow/ragflow/issues/14866)[#14867](https://github.com/infiniflow/ragflow/pull/14867)
- Data source: the GitHub data source connector was failing to sync any content by default. [#13975](https://github.com/infiniflow/ragflow/issues/13975)[#14062](https://github.com/infiniflow/ragflow/pull/14062)
- The Tongyi-Qianwen text embedding models were hitting the wrong API endpoints when configured with international or Chinese regional URLs. [#14784](https://github.com/infiniflow/ragflow/pull/14784)
- Agent: Fully aggregates message content, reference data, and structured outputs across all generated events to fix incomplete responses in the non-streaming `/api/v1/agentbots/<agent_id>/completions` endpoint. [#13384](https://github.com/infiniflow/ragflow/issues/13384)[#14848](https://github.com/infiniflow/ragflow/pull/14848)
- Prevents the **Retrieval** component's manual metadata filters from getting stuck on the first loop's value by making a temporary copy of the filter settings to preserve the original placeholder.[#12582](https://github.com/infiniflow/ragflow/issues/12582)[#14849](https://github.com/infiniflow/ragflow/pull/14849)
- Agent: Fixed MCP tool name duplication.[#14217](https://github.com/infiniflow/ragflow/pull/14217)
- IMAP synchronization process crashed when multiple email addresses or quoted commas were detected in "From" header. [#14963](https://github.com/infiniflow/ragflow/issues/14963)[#14964](https://github.com/infiniflow/ragflow/issues/14964)[#15006](https://github.com/infiniflow/ragflow/pull/15006)
- Langfuse integration failed to track token consumption. [#9837](https://github.com/infiniflow/ragflow/issues/9837)[#13294](https://github.com/infiniflow/ragflow/pull/13294)
- Enhances the stability and fault tolerance of the reranking module by implementing network timeouts, crash-prevention safeguards, and specific provider bug fixes. [#14264](https://github.com/infiniflow/ragflow/pull/14264)
- Increases minimum supported Nginx to 1.31.0. [#14928](https://github.com/infiniflow/ragflow/issues/14928)[#15007](https://github.com/infiniflow/ragflow/pull/15007)
- Introduces a generic RESTful API connector, enabling configurable data ingestion from niche or enterprise-specific platforms into RAGFlow. [#13545](https://github.com/infiniflow/ragflow/pull/13545)
- Agent: Implements tag management to help users categorize, filter, and sort their Agent apps. [#14799](https://github.com/infiniflow/ragflow/pull/14799)
- Adds widget customization and persistence, allowing users to tailor their chat interface and ensure their settings are retained across sessions. [#14603](https://github.com/infiniflow/ragflow/pull/14603)
- Corrects the API endpoint for downloading original files from a specified dataset. [#14910](https://github.com/infiniflow/ragflow/pull/14910) See also [Download document](./references/http_api_reference.md#download-document).
- Enables assigning specific roles like content, metadata, and primary key, to table columns. [#13710](https://github.com/infiniflow/ragflow/pull/13710)
- S3 connector: Implements ETag-based incremental synchronization for S3 data sources, drastically reducing sync time and AWS egress costs for users with massive-volumn S3-based datasets. [#14628](https://github.com/infiniflow/ragflow/issues/14628)[#14677](https://github.com/infiniflow/ragflow/pull/14677)
- Enforces HTTP request timeouts across the LLM integration layer. [#14313](https://github.com/infiniflow/ragflow/pull/14313)
- Refactors thread pool lifecycle management in `file_service.py` and `task_executor.py` for more efficient, lightweight resource handling. [#14668](https://github.com/infiniflow/ragflow/pull/14668)
- Agent: Enables the **Code** component to output and display file-based attachments, such as charts and images, directly in the chat. [#14787](https://github.com/infiniflow/ragflow/pull/14787)
- Images in multi-sheet Excel workbooks were not scoped by sheet, causing images to be incorrectly attributed across different worksheets. [#14120](https://github.com/infiniflow/ragflow/pull/14120)
- Agent: Splits the **Message** component output into distinct 'waiting' and 'message' states when nested inside an **Iteration** component alongside a **Wait** component. [#14839](https://github.com/infiniflow/ragflow/pull/14839)
- Agent: The **Iteration** component failed to correctly pass array elements to its child components due to a naming mismatch between the expected `IterationItem` alias and the runtime `item` variable. [#14146](https://github.com/infiniflow/ragflow/pull/14146)
- Agent: Template strings in tool-type components like **Email** and **Invoke** failed to interpolate; `{{variable}}` placeholders were passed through as raw text. [#14601](https://github.com/infiniflow/ragflow/pull/14601)
- API refactoring and unification: Continues the transition of web APIs to RESTful conventions, ensuring backward compatibility for all legacy endpoints.
### Data source
- Introduces a lightweight snapshot mechanism for synchronizing deleted files across eight data sources—including Moodle, DingTalk AI Table, and RSS—ensuring a faithful reflection of all remote data sources. [#14362](https://github.com/infiniflow/ragflow/issues/14362)[#14499](https://github.com/infiniflow/ragflow/pull/14499)
### Bug fixes
- Metadata visibility issues during v0.24.0 to v0.25.0 upgrades.
- API refactoring and unification: Standardizes web APIs to RESTful conventions across all endpoints, unifying document creation and indexing flows while maintaining backward compatibility.
- Optimizes Chat, Agent, and Search embedded dialog pages for mobile compatibility.
- Underlying system & infrastructure optimization:
- Bumps RAGFlow's document engine, Elasticsearch to 9.x.
- Switches the default object storage container to `pgsty/minio` due to the deprecation of the official MinIO images.
- Adds database migration scripts; see [this readme](https://github.com/infiniflow/ragflow/tree/74b44e1aa3ecd6687b3aa4ef731d0187720c3cb5/tools/scripts) for further details.
- Launches a new Chat-like Agent conversation management interface that retains sessions and dialogue history.
- Introduces a multi-Sandbox mechanism supporting local gVisor and Alibaba Cloud, with compatibility for mainstream Sandbox APIs (configurable in the Admin page).
- Supports exporting Agent outputs in Word or Markdown formats.
- Adds a **List operations** component.
- Adds a **Variable aggregator** component.
- Data sources:
- Supports S3-compatible data sources, e.g., MinIO.
- Adds data synchronization with JIRA.
- Continues the redesign of the **Profile** page layouts.
- Upgrades the Flask web framework from synchronous to asynchronous, increasing concurrency and preventing blocking issues caused when requesting upstream LLM services.
- A v0.22.0 issue: Users failed to parse uploaded files or switch embedding model in a dataset containing parsed files using a built-in model from a `-full` RAGFlow edition.
From this release onwards, we ship only the slim edition (without embedding models) Docker image and no longer append the `-slim` suffix to the image tag.
:::
### New Features
- Dataset:
- Supports data synchronization from five online sources (AWS S3, Google Drive, Notion, Confluence, and Discord).
- RAPTOR can be built across an entire dataset or on individual documents.
- Ingestion pipeline: Supports [Docling document parsing](https://github.com/docling-project/docling) in the **Parser** component.
- Launches a new administrative Web UI dashboard for graphical user management and service status monitoring.
- Agent:
- Supports structured output.
- Supports metadata filtering in the **Retrieval** component.
- Introduces a **Variable aggregator** component with data operation and session variable definition capabilities.
### Improvements
- Agent: Supports visualizing previous components' outputs in the **Await Response** component.
- Revamps the model provider page.
- Upgrades RAGFlow's document engine Infinity to v0.6.5.
### Added Models
- Kimi-K2-Thinking
### New agent templates
- Interactive Agent, incorporates real-time user feedback to dynamically optimize Agent output.
- Orchestratable ingestion pipeline: Supports customized data ingestion and cleansing workflows, enabling users to flexibly design their data flows or directly apply the official data flow templates on the canvas.
- GraphRAG & RAPTOR write process optimized: Replaces the automatic incremental build process with manual batch building, significantly reducing construction overhead.
- Long-context RAG: Automatically generates document-level table of contents (TOC) structures to mitigate context loss caused by inaccurate or excessive chunking, substantially improving retrieval quality. This feature is now available via a TOC extraction template. See [here](./guides/dataset/advanced/extract_table_of_contents.md).
- Video file parsing: Expands the system's multimodal data processing capabilities by supporting video file parsing.
- Admin CLI: Introduces a new command-line tool for system administration, allowing users to manage and monitor RAGFlow's service status via command line.
### Improvements
- Redesigns RAGFlow's Login and Registration pages.
- Company Research Report Deep Dive Agent: Designed for financial institutions to help analysts quickly organize information, generate research reports, and make investment decisions.
- Orchestratable Ingestion Pipeline Template: Allows users to apply this template on the canvas to rapidly establish standardized data ingestion and cleansing processes.
- Four framework-level prompt blocks are available in the **System prompt** section, enabling customization and overriding of prompts at the framework level, thereby enhancing flexibility and control. See [here](./guides/agent/agent_component_reference/agent.mdx#system-prompt).
- **Execute SQL** component enhanced: Replaces the original variable reference component with a text input field, allowing users to write free-form SQL queries and reference variables. See [here](./guides/agent/agent_component_reference/execute_sql.md).
- An Agent in task mode still required a dialogue to trigger.
- Repeated answers in multi-turn dialogues.
- Duplicate summarization of parallel execution results.
### API changes
#### HTTP APIs
- Adds a body parameter `"metadata_condition"` to the [Retrieve chunks](./references/http_api_reference.md#retrieve-chunks) method, enabling metadata-based chunk filtering during retrieval. [#9877](https://github.com/infiniflow/ragflow/pull/9877)
#### Python APIs
- Adds a parameter `metadata_condition` to the [Retrieve chunks](./references/python_api_reference.md#retrieve-chunks) method, enabling metadata-based chunk filtering during retrieval. [#9877](https://github.com/infiniflow/ragflow/pull/9877)
Ecommerce Customer Service Workflow: A template designed to handle enquiries about product features and multi-product comparisons using the internal dataset, as well as to manage installation appointment bookings.
- Documentation: Corrects inaccuracies in the API reference.
### New Agent templates
- Report Agent: A template for generating summary reports in internal question-answering scenarios, supporting the display of tables and formulae. [#9427](https://github.com/infiniflow/ragflow/pull/9427)
- Choose Your Knowledge Base Workflow: Lets users select a dataset to query during conversations. [#9325](https://github.com/infiniflow/ragflow/pull/9325)
- Choose Your Knowledge Base Agent: Delivers higher-quality responses with extended reasoning time, suited for complex queries. [#9325](https://github.com/infiniflow/ragflow/pull/9325)
### Fixed Issues
- The **Agent** component was unable to invoke models installed via vLLM.
- Agents could not be shared with the team.
- Embedding an Agent into a webpage was not functioning properly.
From v0.20.0 onwards, Agents are no longer compatible with earlier versions, and all existing Agents from previous versions must be rebuilt following the upgrade.
### New features
- Unified orchestration of both Agents and Workflows.
- A comprehensive refactor of the Agent, greatly enhancing its capabilities and usability, with support for Multi-Agent configurations, planning and reflection, and visual functionalities.
- Fully implemented MCP functionality, allowing for MCP Server import, Agents functioning as MCP Clients, and RAGFlow itself operating as an MCP Server.
- Multi-Agent based Deep research: Collaborative Agent teamwork led by a Lead Agent with multiple Subagents, distinct from traditional workflow orchestration.
- Added support for models installed via Ollama or VLLM when creating a dataset through the API. [#8069](https://github.com/infiniflow/ragflow/pull/8069)
- [Cross-language search](./references/glossary.mdx#cross-language-search) is supported in the Knowledge and Chat modules, enhancing search accuracy and user experience in multilingual environments, such as in Chinese-English datasets.
- Agent component: A new Code component supports Python and JavaScript scripts, enabling developers to handle more complex tasks like dynamic data processing.
- Enhanced image display: Images in Chat and Search now render directly within responses, rather than as external references. Knowledge retrieval testing can retrieve images directly, instead of texts extracted from images.
From this release onwards, built-in rerank models have been removed because they have minimal impact on retrieval rates but significantly increase retrieval time.
- DeepDoc supports adopting VLM model as a processing pipeline during document layout recognition, enabling in-depth analysis of images in PDF and DOCX files.
- AI chat: Implements Deep research for agentic reasoning. To activate this, enable the **Reasoning** toggle under the **Prompt engine** tab of your chat assistant dialogue.
- AI chat: Leverages Tavily-based web search to enhance contexts in agentic reasoning. To activate this, enter the correct Tavily API key under the **Assistant settings** tab of your chat assistant dialogue.
- Dataset: Adds a **PDF parser**, aka **Document parser**, dropdown menu to dataset configurations. This includes a DeepDoc model option, which is time-consuming, a much faster **naive** option (plain text), which skips DLA (Document Layout Analysis), OCR (Optical Character Recognition), and TSR (Table Structure Recognition) tasks, and several currently *experimental* large model options. See [here](./guides/dataset/select_pdf_parser.md).
- Agent component: **(x)** or a forward slash `/` can be used to insert available keys (variables) in the system prompt field of the **Generate** or **Template** component.
- Object storage: Supports using Aliyun OSS (Object Storage Service) as a file storage option.
- GraphRAG refactor: Knowledge graph is dynamically built on an entire dataset rather than on an individual file, and automatically updated when a newly uploaded file starts parsing. See [here](./guides/dataset/advanced/construct_knowledge_graph.md).
- Adds an **Iteration** agent component and a **Research report generator** agent template. See [here](./guides/agent/agent_component_reference/iteration.mdx).
Adds [Infinity's configuration file](https://github.com/infiniflow/ragflow/blob/main/docker/infinity_conf.toml) to facilitate integration and customization of [Infinity](https://github.com/infiniflow/infinity) as a document engine. From this release onwards, updates to Infinity's configuration can be made directly within RAGFlow and will take effect immediately after restarting RAGFlow using `docker compose`. [#3715](https://github.com/infiniflow/ragflow/pull/3715)
- Supports [Infinity](https://github.com/infiniflow/infinity) or Elasticsearch (default) as document engine for vector storage and full-text indexing. [#2894](https://github.com/infiniflow/ragflow/pull/2894)
From this release onwards, **service_config.yaml.template** replaces **service_config.yaml** for configuring backend services. Upon Docker container startup, the environment variables defined in this template file are automatically populated and a **service_config.yaml** is auto-generated from it. [#3341](https://github.com/infiniflow/ragflow/pull/3341)
This approach eliminates the need to manually update **service_config.yaml** after making changes to **.env**, facilitating dynamic environment configurations.
Ensure that you [upgrade **both** your code **and** Docker image to this release](./administrator/upgrade_ragflow.mdx#upgrade-ragflow-to-the-most-recent-officially-published-release) before trying this new approach.
- Implements an **Excel to HTML** toggle in the **General** chunking method, allowing users to parse a spreadsheet into either HTML tables or key-value pairs by row.
From this release onwards, RAGFlow offers slim editions of its Docker images to improve the experience for users with limited Internet access. A slim edition of RAGFlow's Docker image does not include built-in BGE/BCE embedding models and has a size of about 1GB; a full edition of RAGFlow is approximately 9GB and includes two built-in embedding models.
- Integrates reranker and embedding models: [BCE](https://github.com/netease-youdao/BCEmbedding), [BGE](https://github.com/FlagOpen/FlagEmbedding), and [Jina](https://jina.ai/embeddings/).
- Supports LLMs Baichuan and VolcanoArk.
- Implements [RAPTOR](https://arxiv.org/html/2401.18059v1) for improved text retrieval.
