FastGPT v4.16.0-beta2
FastGPT v4.16.0-beta2
📦 升级指南
1. 清洗历史系统模型配置
本版本开始在系统模型初始化及保存时使用严格 Schema。此前版本保存的数字字符串、字符串形式的价格梯度或缺失字段可能导致初始化校验失败。升级后请先执行 dry-run,查看需要处理的模型配置;dry-run 不会修改数据或刷新缓存:
curl -X POST 'https://你的域名/api/admin/dataClean/cleanSystemModelConfigs' \
-H 'Content-Type: application/json' \
-H 'rootkey: 你的ROOT_KEY' \
-d '{"dryRun":true}'
确认 invalidSamples 中没有需要人工处理的数据后,执行正式清洗:
curl -X POST 'https://你的域名/api/admin/dataClean/cleanSystemModelConfigs' \
-H 'Content-Type: application/json' \
-H 'rootkey: 你的ROOT_KEY' \
-d '{"dryRun":false}'
清洗会将合法数字字符串转换为 number、将字符串形式的 priceTiers 转换为数组,并删除非法的可选数字。非法或缺失的必填数字使用系统默认值:LLM 的 maxContext/maxResponse/quoteMaxToken 分别为 16000/16000/13000,Embedding 的 defaultToken/maxToken 分别为 500/3000,价格为 0。functionCall 保持可选,Embedding 缺失的 weight 补为 0。
正式执行会统一写入数据并立即刷新系统模型缓存;即使没有记录需要更新,也会重新构建运行时缓存。接口可安全重复执行;再次 dry-run 时,wouldUpdate 应为 0。无法通过当前完整模型 Schema 的记录不会写入,详情会全部返回在 invalidSamples 中。
🚀 新增内容
- 工作流应用的系统配置移至画布左侧工具栏中的独立配置面板,新建工作流应用时会自动打开。
- 开场白支持独立配置多个预设问题,并可拖拽排序。
⚙️ 优化
- 重构发布渠道页面,按原生渠道和第三方渠道分组展示,并显示各渠道已配置数量。
- 工具市场批量更新支持展示部分失败项、单独重试或卸载,并完善已安装版本和更新状态的展示。
- 门户页快捷应用数量上限调整为 3 个,并兼容已超出上限的历史配置。
- PDF 解析器动态边缘裁剪,而不是固定边缘裁剪,避免裁剪掉真实内容。
🐛 修复
- 修复旧版工作流 HTTP 工具的动态参数未正确恢复默认输入方式的问题。
- 修复新建工作流应用时,国际化资源未预加载导致初始配置文案异常的问题。
- 修复工作流工具参数可能被重复渲染的问题。
- 修复应用发布后,文件变量无法上传文件的问题。
- 修复共享工作流工具的文件参数无法上传文件的问题。
- 修复 S3 对象键和文件名包含空格、
%、#、?、斜杠等特殊字符时,可能导致上传、解析、预览或下载异常的问题。 - 修复系统默认模型未进行敏感信息过滤的问题,避免系统初始化接口响应返回模型 API Key、请求地址及内部配置。
🛠️ 代码优化
- 审计日志归档,不再删除,改成转存到 S3 冷归档。
- 增加对 admin 配置的数据校验和清洗。
- 取消 account/* 页面 SSR。
What’s Changed
- feat(tool-marketplace): support partial plugin update retries by @FinleyGe in https://github.com/labring/FastGPT/pull/7444
- fix: allow shared workflow tool file uploads by @xqvvu in https://github.com/labring/FastGPT/pull/7472
- Doc by @c121914yu in https://github.com/labring/FastGPT/pull/7471
- feat(outlink): new configuration page by @Nixieboluo in https://github.com/labring/FastGPT/pull/7456
- docs: add 4.16.0-beta2 publish channel notes by @c121914yu in https://github.com/labring/FastGPT/pull/7476
- fix(docs): include SWC ESM helpers in standalone output by @c121914yu in https://github.com/labring/FastGPT/pull/7481
- ci(docs): publish deployment images to GHCR by @c121914yu in https://github.com/labring/FastGPT/pull/7482
- chore(helm): move charts and release workflow to fastgpt-pro by @DigHuang in https://github.com/labring/FastGPT/pull/7463
- fix: localize invalid coupon errors by @c121914yu in https://github.com/labring/FastGPT/pull/7483
- docs(upgrade): add v4.14.30 upgrade guide by @FinleyGe in https://github.com/labring/FastGPT/pull/7484
- sanitize s3 key by @xqvvu in https://github.com/labring/FastGPT/pull/7468
- fix(agent): use app owner tmbId for sandbox skill authentication by @DigHuang in https://github.com/labring/FastGPT/pull/7486
- feat(app): move system settings configuration by @DigHuang in https://github.com/labring/FastGPT/pull/7454
- docs: update v4.16.0-beta2 release notes by @c121914yu in https://github.com/labring/FastGPT/pull/7489
- feat: deduplicate CRM lifecycle reports by @c121914yu in https://github.com/labring/FastGPT/pull/7490
- fix(global): add .js extension to ajv ESM subpath imports by @ctlaltlaltc in https://github.com/labring/FastGPT/pull/7492
- fix(app): make dataset search configurable in agent v2 by @YYH211 in https://github.com/labring/FastGPT/pull/7494
- Improve API documentation sidebar styling by @c121914yu in https://github.com/labring/FastGPT/pull/7493
- fix: align Zod contracts with runtime data by @c121914yu in https://github.com/labring/FastGPT/pull/7499
- fix(app): show image recognition for vision-only models by @YYH211 in https://github.com/labring/FastGPT/pull/7501
- fix(workflow): node debug file upload uses draft config and session chatId by @xqvvu in https://github.com/labring/FastGPT/pull/7504
- fix(chat): standardize preview layout and content width by @DigHuang in https://github.com/labring/FastGPT/pull/7502
- fix(workflow): project external variables as inputs by @FinleyGe in https://github.com/labring/FastGPT/pull/7506
- fix: polish app modals and configuration handling by @c121914yu in https://github.com/labring/FastGPT/pull/7505
- fix: fallback tool input descriptions by @c121914yu in https://github.com/labring/FastGPT/pull/7508
- fix(workflow): preserve explicit external variable input modes by @c121914yu in https://github.com/labring/FastGPT/pull/7510
- fix(docs): handle missing static assets correctly by @c121914yu in https://github.com/labring/FastGPT/pull/7511
- fix(sandbox): use Cilium isolation for Kubernetes OpenSandbox by @DigHuang in https://github.com/labring/FastGPT/pull/7514
- refactor(sandbox): OpenSandbox stream handling and replace with filesystem API by @DigHuang in https://github.com/labring/FastGPT/pull/7500
- fix: do not treat reasoning-only tool_calls finish as empty response by @LHMQ878 in https://github.com/labring/FastGPT/pull/7517
- fix(s3): allow deleting legacy object keys rejected by validation by @xqvvu in https://github.com/labring/FastGPT/pull/7509
- docs: fix broken star history chart in READMEs by @Dessalines39394 in https://github.com/labring/FastGPT/pull/7518
- refactor: migrate account pages to client i18n by @c121914yu in https://github.com/labring/FastGPT/pull/7515
- fix: sanitize default models in system init data by @c121914yu in https://github.com/labring/FastGPT/pull/7519
- perf: publish multi-arch images directly to aliyun by @c121914yu in https://github.com/labring/FastGPT/pull/7521
New Contributors
- @LHMQ878 made their first contribution in https://github.com/labring/FastGPT/pull/7517
- @Dessalines39394 made their first contribution in https://github.com/labring/FastGPT/pull/7518
Full Changelog: https://github.com/labring/FastGPT/compare/v4.16.0-beta1…v4.16.0-beta2
Source labring/FastGPT ·
v4.16.0-beta2· 2026-08-14 · upstream release
Mirrored verbatim for indexing. All rights belong to the upstream project.
Release notes are published by the upstream project and mirrored here for indexing. All rights belong to their authors.