This article records configuration and changes from a historical FastGPT release. Use it to understand that release, and consult the release notes for your installed and target versions before applying dependencies, configuration, or migration steps.
Key Feature Additions
This release includes multiple core feature expansions:
- A multi-stage search pipeline with separated vector semantic search, full-text search, and reranking modes, using Reciprocal Rank Fusion (RRF) to merge final results.
- The HTTP node now supports a JSON editor for configuring request headers.
- Official support for ReRank model deployment, with a dedicated setup guide.
A summarized breakdown of new features is below:
| Feature Category | Update Details | Reference Link |
|---|---|---|
| Search Workflow | Three distinct search modes with RRF result merging | N/A |
| HTTP Node Configuration | JSON editor for request header setup | N/A |
| ReRank Model Support | Official deployment documentation available | ../../custom-models/bge-rerank.en.mdx |
Operational and Code Improvements
Several quality-of-life and bug fix updates are included:
- Question classifier prompts have been updated to use ID-guided classification, and validated to work with Chinese commercial API models including Baidu, Alibaba, Zhipu, and iFlytek when using Prompt mode.
- Codebase optimization: Application icons have been extracted and auto-generated to reduce maintenance overhead.
- A critical bug for link-based Collections has been resolved: previously, saved selectors were not stored, causing sync operations to run without configured selectors.
UI and Strategic Updates
The FastGPT user interface will begin rolling out a new design gradually across the platform. For details on upcoming planned changes, refer to the official FastGPT 2024 RoadMap: https://github.com/labring/FastGPT?tab=readme-ov-file#-%E5%9C%A8%E7%BA%BF%E4%BD%BF%E7%94%A8