chore(release): prepare v2.11.6
This commit is contained in:
1 parent
d7cde6c2fc
commit
971d007de7
6 files changed
+103
-3
No files matched your search
@@ -2,6 +2,10 @@
|
||||
|
||||
Full release narratives now live in versioned files under `releases/`. This file stays as the index and summary entry point.
|
||||
|
||||
## [2.11.6] - 2026-06-09
|
||||
- EN: This patch updates the MiniMax default model to M3, fixes OpenAI large image-model parameter handling, supports Unicode variable names, and improves provider error details. See [Release Notes (EN)](releases/v2.11.6.en.md).
|
||||
- 中文:本次补丁将 MiniMax 默认模型更新到 M3,修复 OpenAI 大图模型参数兼容,支持 Unicode 变量名,并优化 provider 错误详情。参见 [版本说明(中文)](releases/v2.11.6.zh-CN.md)。
|
||||
|
||||
## [2.11.5] - 2026-05-28
|
||||
- EN: This patch adds Xiaomi MiMo Token Plan provider support, hardens malformed text-model configuration handling, expands deployment guidance, and tidies deployment/project scaffolding. See [Release Notes (EN)](releases/v2.11.5.en.md).
|
||||
- 中文:本次补丁新增 Xiaomi MiMo Token Plan provider,强化异常文本模型配置处理,扩展部署说明,并整理部署与项目脚手架。参见 [版本说明(中文)](releases/v2.11.5.zh-CN.md)。
|
||||
|
||||
+1
-1
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "prompt-optimizer",
|
||||
"version": "2.11.5",
|
||||
"version": "2.11.6",
|
||||
"private": true,
|
||||
"packageManager": "pnpm@10.6.1",
|
||||
"engines": {
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "@prompt-optimizer/desktop",
|
||||
"version": "2.11.5",
|
||||
"version": "2.11.6",
|
||||
"description": "Desktop application for Prompt Optimizer",
|
||||
"main": "main.js",
|
||||
"repository": {
|
||||
|
||||
@@ -2,7 +2,7 @@
|
||||
"manifest_version": 3,
|
||||
"default_locale": "zh_CN",
|
||||
"name": "__MSG_extName__",
|
||||
"version": "2.11.5",
|
||||
"version": "2.11.6",
|
||||
"description": "__MSG_extDesc__",
|
||||
"icons": {
|
||||
"16": "icons/icon16.png",
|
||||
|
||||
@@ -0,0 +1,48 @@
|
||||
# Prompt Optimizer v2.11.6
|
||||
|
||||
## Summary
|
||||
- Updates the built-in MiniMax text provider lineup so MiniMax-M3 is the default while MiniMax-M2.7 variants remain selectable.
|
||||
- Fixes OpenAI large image-model calls by avoiding the unsupported `response_format` parameter when the selected model does not support it.
|
||||
- Allows Unicode variable names in the variable manager/editor, including Chinese variable identifiers.
|
||||
- Improves provider/evaluation error formatting so detailed API errors are easier to read and diagnose.
|
||||
|
||||
## Highlights
|
||||
- Model and prompt workflows are a little more forgiving: newer MiniMax defaults are available immediately, non-English variable names work in the manager, and provider failures surface clearer details.
|
||||
|
||||
## Product Updates
|
||||
### Desktop
|
||||
- Desktop version metadata is synchronized to v2.11.6 and inherits the shared MiniMax, image-provider, variable-manager, and error-formatting fixes.
|
||||
### Web
|
||||
- MiniMax-M3 is now the default MiniMax text model, with MiniMax-M2.7 and MiniMax-M2.7-highspeed still available as alternatives.
|
||||
- Variable management now accepts Unicode names consistently across the editor and manager modal.
|
||||
- Provider/evaluation error messages now preserve and format structured detail so users can see the actionable API failure instead of a flattened generic message.
|
||||
### Extension
|
||||
- Extension version metadata is synchronized to v2.11.6 and inherits the shared provider and variable-handling updates.
|
||||
### Core/Infra
|
||||
- Removed deprecated MiniMax-M2.5 variants from the built-in MiniMax model list and updated adapter tests for the new lineup.
|
||||
- Updated OpenAI image adapter parameter handling and regression coverage for large image models that reject `response_format`.
|
||||
- Added regression coverage for Unicode variable names and formatted provider error details.
|
||||
|
||||
## Fixes
|
||||
- Fixed OpenAI large image-model requests failing with `Unknown parameter: 'response_format'`.
|
||||
- Fixed variable manager validation rejecting Unicode variable names such as Chinese identifiers.
|
||||
- Fixed provider/evaluation error details being hard to inspect after API failures.
|
||||
|
||||
## Breaking Changes / Upgrade Notes
|
||||
- None.
|
||||
|
||||
## Developer Notes
|
||||
- The release range is `v2.11.5..HEAD`.
|
||||
- Version sync updated the root package, desktop package, and extension manifest to `2.11.6`.
|
||||
- Verification covered `pnpm test:gate`, `node scripts/release-notes.js check v2.11.6`, and `git diff --check`.
|
||||
|
||||
<!-- Release drafting reference:
|
||||
Remove every TODO placeholder before running "pnpm release:notes:check v2.11.6".
|
||||
|
||||
- Range: v2.11.5..HEAD
|
||||
- Merge pull request #308 from chaozai0304/develop (d7cde6c2)
|
||||
- fix(evaluation): format provider error details (ecbe7889)
|
||||
- fix(ui): allow unicode variable names in manager (e0392578)
|
||||
- Fixed the OpenAI error when calling large image models: Unknown parameter: 'response_format'. (request id: 202606030925559064897917EvHxr7G) (9ed01ce2)
|
||||
- feat: upgrade MiniMax default model to M3 (#307) (74b06cea)
|
||||
-->
|
||||
@@ -0,0 +1,48 @@
|
||||
# Prompt Optimizer v2.11.6
|
||||
|
||||
## 概括
|
||||
- 更新内置 MiniMax 文本 provider 模型列表,将 MiniMax-M3 设为默认模型,同时保留 MiniMax-M2.7 系列可选项。
|
||||
- 修复 OpenAI 大图模型调用时携带不支持的 `response_format` 参数导致失败的问题。
|
||||
- 变量管理器 / 编辑器现在支持 Unicode 变量名,包括中文变量标识符。
|
||||
- 优化 provider / 评估错误详情格式化,让 API 失败原因更容易阅读和定位。
|
||||
|
||||
## 亮点
|
||||
- 模型与提示词工作流更顺手:MiniMax 默认模型更新到 M3,非英文变量名可以直接管理,provider 失败时也能看到更清楚的错误详情。
|
||||
|
||||
## 产品更新
|
||||
### Desktop
|
||||
- 桌面端版本元数据已同步到 v2.11.6,并继承共享的 MiniMax、图像 provider、变量管理和错误格式化修复。
|
||||
### Web
|
||||
- MiniMax-M3 现在是默认 MiniMax 文本模型,MiniMax-M2.7 和 MiniMax-M2.7-highspeed 仍可作为备选模型使用。
|
||||
- 变量编辑器与变量管理弹窗现在一致支持 Unicode 变量名。
|
||||
- Provider / 评估错误消息会保留并格式化结构化详情,用户能看到更可操作的 API 失败信息,而不是被压平的泛化提示。
|
||||
### Extension
|
||||
- 扩展端版本元数据已同步到 v2.11.6,并继承共享的 provider 与变量处理更新。
|
||||
### Core/Infra
|
||||
- 从内置 MiniMax 模型列表中移除过期的 MiniMax-M2.5 系列,并更新 adapter 测试覆盖新的模型组合。
|
||||
- 更新 OpenAI 图像 adapter 参数处理,并补充大图模型拒绝 `response_format` 的回归覆盖。
|
||||
- 补充 Unicode 变量名和 provider 错误详情格式化的回归测试。
|
||||
|
||||
## 修复
|
||||
- 修复 OpenAI 大图模型请求因 `Unknown parameter: 'response_format'` 失败的问题。
|
||||
- 修复变量管理器校验拒绝中文等 Unicode 变量名的问题。
|
||||
- 修复 provider / 评估 API 失败后的错误详情不便检查的问题。
|
||||
|
||||
## 破坏性变更 / 升级说明
|
||||
- 无。
|
||||
|
||||
## 开发者说明
|
||||
- 本次发布范围是 `v2.11.5..HEAD`。
|
||||
- 版本同步已将根包、桌面包和扩展 manifest 更新到 `2.11.6`。
|
||||
- 验证覆盖 `pnpm test:gate`、`node scripts/release-notes.js check v2.11.6` 和 `git diff --check`。
|
||||
|
||||
<!-- Release drafting reference:
|
||||
Remove every TODO placeholder before running "pnpm release:notes:check v2.11.6".
|
||||
|
||||
- Range: v2.11.5..HEAD
|
||||
- Merge pull request #308 from chaozai0304/develop (d7cde6c2)
|
||||
- fix(evaluation): format provider error details (ecbe7889)
|
||||
- fix(ui): allow unicode variable names in manager (e0392578)
|
||||
- Fixed the OpenAI error when calling large image models: Unknown parameter: 'response_format'. (request id: 202606030925559064897917EvHxr7G) (9ed01ce2)
|
||||
- feat: upgrade MiniMax default model to M3 (#307) (74b06cea)
|
||||
-->
|
||||
Reference in new issue
Block a user