Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r619 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.10.1.570...031c56f

Commits between 1.10.1.570 and 031c56f8cbbb8eb0bf57f4b7abca402d4e5a4feb:

• Merge pull request #123 from sdcb/feature/1.10.2
• add 1.10.2 release notes
• feat(admin-dashboard): switch summary cards to range-based chat and active user stats
• fix(fe): move collapsed user message toggle below text
• feat(chat): embed user message expand/collapse control into the message bubble with a 5.5-line preview
• fix: remove fixed max-height cap from ThinkingMessage expansion
• feat(request-trace): refine compare dialog and column selection UX
• feat(request-trace): refine trace UI, details loading, and quick presets
• fix(api): keep all Anthropic tool results when converting to OpenAI payloads
• fix(RequestTracePersistService): remove redundant delete operation for RequestTracePayloads
• feat(request-tracing): record requests earlier and fix observer body capture
• fix(be): capture 401 inbound request traces by reordering middleware
• fix: don't clamp if reasoningEffortId is 0, which indicates that reasoning effort is not specified and the system should use the default
• fix(RequestTraceDetailsDialog): enhance overflow handling and scrollbar styles
• feat(request-trace) save json as formatted
• Fix deepseek api anthropic protocol thinking signature issue
• fix: improve request trace redaction defaults and ordering
• refactor(request-trace): remove redundant description for JSON fields in captured body
• feat(request-trace): add configurable URL parameter redaction
• Refine request trace gateway presets by source and URL pattern
• feat(request-trace): change request trace sample rate to a percentage slider and add localized inline guidance
• refactor(request-trace): unify include/exclude filters and simplify config dialog
• feat(request-trace): reorder direction filter, unify control widths, clarify label, and use tilde-separated column params
• feat(request-trace): add exclude URL patterns to quick presets and enhance UsageRecordsTab with export tips
• feat(request-trace): add retention lifecycle and body-length tracing
• update wechat qrcode
• feat(RequestTracing): integrate request tracing across multiple services and update localization for new fields
• feat(admin): move request trace config to request-trace page and enhance config dialog UX
• feat(FE): update request trace ID type from number to string across multiple components
• feat(RequestTraceQueue): add configuration options for in-memory queue capacity
• refactor(request-tracing): switch to Guid-linked updates, improve queue observability, and short-circuit tracing on entry drop
• feat: add Request Trace functionality with details and export options
• feat(OutboundRequestTraceHandler): add trace ID to outbound request tracing
• feat(RequestTrace): add exception handling and logging to request tracing middleware and persist service
• feat(RequestTracing): add capture streams and middleware for enhanced request/response tracing
• feat(RequestTrace): refactor RequestTrace schema and middleware for enhanced tracking timestamps
• feat(RequestTrace): integrate IUrlEncryptionService for user ID decryption in middleware and handler
• fix(RequestTraceHelper): update MatchesPatterns to handle empty patterns result
• RequestTraceLog Split into 4 pharse save
• Implementing backend tracking infra
• chore(fe): upgrade Next.js to 16.1.6 and remove direct baseline-browser-mapping dev dependency
• Update wechat qrcode link
• maxTextCharsForTruncate
• change into MaxTextCharsForTruncate
• Implement RequestTrack frontend config
• SourceNames -> SourcePatterns
• Add RequestTraceConfig/RequestTrace infra db design
• [fix(mcp): preserve readable Unicode in fetched tool input schemas by replacing GetRawText() with relaxed JSON serialization](h...

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r570 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.10.0.522...c5fb449

Commits between 1.10.0.522 and c5fb449e3f0b77f15b1dabe31133f930e15a043d:

• Merge pull request #121 from sdcb/feature/1.10.1
• update docs
• create latest manifest for non main branch
• Increase default MaxOutputBytes from 6K to 8K, default code-interpreter image to latest
• add rg to code-interpreter image
• add run version in code-interpreter skills.md
• feat(fe): remove RMB symbols from cost displays and simplify price formatting
• feat(sandbox-manager): revamp session info UI, add touch refresh flow, and improve env var editor UX
• Refactor think-tag parsing to use ChatSegment stream and simplify ChatService integration
• feat(cache): unify and optimize ETag handling across models and prompts APIs
• feat(messages): add server-side ETag caching with stable hash for GET /api/messages/{chatId}
• feat(cache): switch chat groups cache to server-driven ETag and remove frontend localStorage cache
• refactor(generate-info): unify admin/user turn-step endpoints via messages API and remove admin-specific turn info path
• Fix(admin/messages) user query filter debounce issue
• ALiyun Oss also supports download as original file name
• Fix FE compilation error caused by recent change
• Show entire api key after create using dialog, because some user can't copy by click copy button
• make file service add button UI consistant with model management UI
• Optimize file service design/user exp, add validate button & frontend mask
• Upgrade baseline-browser-mapping to 2.9.19
• Fix local file service download content disposition
• Add FileName to download requests and set Content-Disposition for S3/Azure/local downloads
• Update WeChat group QR code image link
• Fix overlay with ChatInput bug caused by height in ChatMessage
• Fix Maximum update depth exceeded when refresh page
• UserMessage/ResponseMessage delete icon show loading state when request in progress
• Remove all 3 requestAnimationFrame in ChatView
• Add some auto space in ChatMessage bottom, ensure auto scroll user message to top when send message/select chat
• Add little bit color differnt in ChatInput
• feat: Introduce ChatMiniMap
• refact: Move ChatInput's file upload buttons into + button
• fix: fix send button in ChatInput to the right(regression)
• fix ChatInput height overwrite ChatMessage issue(regression)
• refactor: Add IHttpClientFactory to ImageChatSegment
• refactor: All backend HttpClient use injected IHttpClientFactory
• Update WeChat group QR code image link
• Update WeChat group QR code image URL
• Fix unit test(caused by recent docker interface refactor
• Move agent link to bottom ChatInput
• Optimize docker manager looks and feels
• Optimize env management copy in docker manager
• Extract env variable management as a individual tab in session manage window
• Introduce environment variable management in docker session window
• Optimize pull image display message
• Don't do extra log because progress already yield in EnsureDocker
• Fix the bug causing file list double load
• Keep state after tab switch in session manager window
• Use same error layout between Command execute & File explorer(in docker manage window)
• Fix the bug that session not terminated in db after destroy_session
• Show progress when pulling docker
• Move ListDirectoryAsync out of DockerService as interface layer
• [Delete stupid WrapDockerOperationAsync in docker service](4e174ddcdd...

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r522 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.10.0.520...6d60ad5

Commits between 1.10.0.520 and 6d60ad53aea82e4d2d731f2feda0819da56c376e:

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r416 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.9.0.357...1e98b26

Commits between 1.9.0.357 and 1e98b26935b7ae82a4ce4138e1f53a137939ab76:

• Fix chat competion api cache issue
• fix CS0114 caused by googleai2 chat service
• add 1.9.1 release notes
• Image default wider edge to 1536 instead of 1792, remove useless enums
• fix issue that step info not showing in mobile
• add introduce date to all model providers
• Optimize api key format in usage page
• Enforce one truth from url for usage page
• parallel_tool_calls is only allowed when 'tools' are specified
• Xiaomi use self hosted icon
• don't need api-key header in replace for authorization bearer for mimo
• redact an unmasked api key
• Add UserAgent for all chats httpclient usage
• latex formula now supports copy by introduce copy hook
• fix http 500 issue for {turnId}/{contentId}/text-and-save-new method
• Change all dump into .dump, fix mimo model think issue, refactor dump test to reflect correct chunk
• discard gemini signature when no tool call(better support gemini 3)
• Fix google ai list models
• npm audit fix(upgrade next.js version)
• fix support to gpt-image-1.5 model
• fix generated image always rectangle issue
• Fix non-streaming api call issue when response tool property is null(for xiaomi-mimo)
• Add Xiaomi Mimo model provider(id=21)
• Thinking message time duration use time from step
• Delete StepDivider and use generate info bubble for step info usage
• Introduce chat/turn/step/content 4 level chatview
• RemoveNonCurrentTurnThinkingBlocks in chatservice preprocess
• Auto scroll to leafMessage when messages load
• update leafMessageId to cache, fix not showing correct leaf message when refreshed
• Add dump request assert
• Support interleaved thinking for minimax/deepseek in chat completions api
• detect first and then execute
• fix 'The requested operation requires an element of type 'String', but the target element has type 'Object'. caused by last commit
• Optimize error in non async model for response api
• Show error entire json instead of just message when response api failed
• Consolidate ChatService exception as RawChatServiceException/ChatServiceException only
• Fix FiddlerHttpDump tests
• Fix prompt token not counted issue when user stopped chat, fix response api tool call issue, caused by recent openai .net sdk removal
• TokenPony model provider now supports thinking for deepseek-v3.* models
• Add google gemini unit test
• Make subtitle font smaller for small screen like phone in login page
• Fix fe build warning caused by _error.tsx
• Optimize cache trigger in CI
• Add back env
• Move version out of window global var but as an env variable
• update some other patch packages
• update other patch only FE packages
• chore: update all @radix-ui packages to latest versions
• Fix strange ChatModelDropdownMenu flash issues that caused by next.js/react upgrade
• Fix anthropic first chat fail issue
• Harden usage-source aware chat processing
• Fix a gemini stop text empty issue
• Confirmed gemini tool call worked
• [Remove Mscc.GenerativeAI, ...

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r357 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.8.1.279...32a066c

Commits between 1.8.1.279 and 32a066c3550ca0b011c3739da855939694aef2f3:

• Remove 1809 docker and update readme
• Merge branch 'dev' of github.com:sdcb/chats into dev
• Ensure .NET 10 SDK before build
• Fix the tool call issue in chat completion service, with integration test fix
• Fix a DBUpdate issue when get or insert ClientInfo/ClientIP/UserAgent
• Optimize readme
• Enhance README with badges and formatting updates
• Enhance README with badges and formatting updates
• Fix the StepContentThink delete is not cascade issue
• Add 1.9.0 release notes
• add missing translate for docs
• split docs page into openai/anthropic
• Add 3 build page for better api development: /build/api-keys/, /build/docs/build/usage
• Use same UserMenuPopover componanent in both chat/admin page, and add animation
• Mask api key for api key page(protect from screenshot etc)
• Fix unable to apply thinking budget issue
• Use JsonPolymorphic for AnthropicStreamEvent
• Fix cornor cases of AnthropicChatService
• Reimplement AnthropicChatService by raw HttpClient, remove Anthropic .NET SDK for some issues
• Disable Google Gemini retry to workaround upstream bug (mscraftsman/generative-ai#144)
• Fix ChatInput height not resetting after sending message, add more debug flag in Test2ChatService
• Fix another strange sub pixel line under UserMessage when user inputs base64 very long text
• fix strange sub pixel line under UserMessage issue
• Optimize Anthropic internal web_search tool frontend rendering, optimize ToolCallBlock look and feel in white theme
• implement openai ImageGeneration/image edit api
• Show FilePreview inside ChatInput, fix the FilePreview is missing in non full screen chatinput mode issue
• Make chat service not inherit IDisposable
• Implement anthropic count tokens api
• Fix system prompt parse in anthropic messages api for claude code support
• feat: enforce ApiType in openai/anthropic controller
• chore: move openai/anthropic controller related stuffs into api folder
• fix anthropic compatible api toolcall finish reason
• implmement anthropic compatible messages api
• Add animation to ChatInput when collapse/expand
• Delete unused const variable in ChatInput
• Add full screen animation to ChatInput
• Fix a chatinput flash issue when chat
• Fix Quick add model failed because recent allowVisionLink change
• Fix ToolCallBlock always auto auto expand issue
• Optimize user initial test model use experience
• Switch ApiType in ModelModal should not impact model price
• Anthropic model supports vision link by default
• Supports claude-opus-4.5 compatibility
• Increase anthropic model defaults when create
• Add Ask DeepWiki badge
• Only show web search/code execute when model really supports it
• Confirmed anthropic web_search & mcp call working
• Anthropic Chat Service try implment internal web search tool
• Add Postgres back since it's supported EF 10 now, upgrade .NET packages
• Fix the issue that gemini can't talk with thinking
• optimize token pony(id=19) model provider icon in dark theme
• Add Anthropic(id=20) model provider
• [Fixed admin mod...

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r279 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.8.1.278...7ab4d6a

Commits between 1.8.1.278 and 7ab4d6aab4f3870372e2e756526aeaa4213d9ac1:

• fix windows docker build
• Update readme build doc link
• Update readme model provider list
• Update 1.8.1 readme

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r278 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.8.0.247...7ab4d6a

Commits between 1.8.0.247 and 7ab4d6aab4f3870372e2e756526aeaa4213d9ac1:

• Merge branch 'dev'
• Fix VS Code Github Copilot openai chat completion api compatible issue
• Fix ChatHeader layout issue when model not available
• Correct future version plan in 1.8.1 release notes
• Correct future version plan in 1.8.1 release notes
• Correct release notes in image type support part
• add 1.8.1 release notes
• feat: Add per-reasoning segment lifecycle tracking and auto-collapse
• ToolCallMessage can collapse, add animation to ToolCall/Thinking UI
• Optimize logic to get user list in admin user-model page
• Optimize admin models page UI
• Add Model->Model User management in admin model page
• Update ByUserTab commit to reflect real situation
• Remove ByModel tab in new user model permission page
• refactor: use URL as single source of truth for user-models filters
• Optimize UserModelPermission ByUser page
• fix compilation error caused by previous user model permission page
• optimize admin user model permission page
• admin user model page initial commit
• fix: add scrolling support to user models drawer
• feat: Add 1.8.1 database migration script for Azure AI Foundry URL cleanup
• feat: Add TokenPony provider (ID=19) and refactor Azure/ChatService architecture
• feat: add loading state to new chat creation button
• feat: add chat list caching and parallel API loading
• fix: Ensure back button will back to home
• Use icon button for theme/language settings
• refactor: replace model provider dropdown with responsive icon grid
• disable openai client retry
• Move supported content type into ChatCompletionService layer, fix gemini-2.0-flash-exp
• Refactor file preview with new FilePreview component
• Move file processing from ChatController to ChatService
• Refactor file handling and remove unused image size tables
• Fix reasoning content extraction/decode issues caused by OpenAI 2.6.0 upgrade
• optimize OpenRouter error message
• Fix issue for reasoning_content can't read after upgrade to OpenAI 2.6 by using self-compiled Sdcb.OpenAI
• Merge branch 'dev' into feature/1.8.1
• road to upgrade to openai 2.6

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r247 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.7.2.209...1.8.0.247

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r209 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.7.1.189...c7237bf

Commits between 1.7.1.189 and c7237bf1ccc03c8645c103f4f75eb423a46147d7:

• Merge branch 'dev'
• update 1.7.2 release notes
• Merge pull request #104 from sangyuxiaowu/patch-1
• fix chatinput css
• add 1.7.2 release notes
• FE support image size adjustments for gpt-image-1-mini model
• fix reasoning effort for api image generation service not support issue
• Introduce Chats.BE.ApiTest, a xunit test project to integrate test openai compatible chat completion api
• Move out chatApi.ts out of Chat.tsx(1431->1384 lines)
• auto route image generation to non streaming if n>1 to fix stream mode not support multiple image
• fix FE build error caused by recent change
• tested FE confirmed streamed ImageGeneration service worked
• enable FE support image streaming function
• clear debug info for ImageGenerationService's streaming function
• remove unnecessary server response(12) when image generation/edit to save tranfic
• add support for streaming gpt-image-1/gpt-image-1-mini image generation/edit api(with some debug info
• fix reasoning effort not take effect issue
• add gpt-5-codex/gpt-5-pro/gpt-image-1-mini in ModelReference
• fix BE compilation error caused by mcp upgrade
• Show tool call instead of raw text output for gemini code execution
• upgrade Mscc.GenerativeAI/AWSSDK.S3 dependency
• add a document to brief how to add a field to ChatConfig
• fix compilation error caused by previous commit
• Introduce CodeExecutionEnabled in ChatConfig
• add reasoning effort minimal for gpt-5
• Add CodeExecutionEnabled field in ChatConfig
• Fix SystemPrompt textarea scrolling bug
• Fix excessive spacing below chat messages after reducing input rows
• bump version to 1.7.2, upgrade 11 third party deps
• fix(fe/markdown): avoid extra blank lines and honor soft break by enable remark-breaks in ResponseMessage

Docker

Alternative binaries download links that may faster than GitHub(for China users)

NOTE:
Replace r189 with latest in the download link to get the latest version, for example: https://chats.sdcb.pub/release/latest/chats-win-x64.zip

Full Changelogs

1.7.0.154...8183a60

Commits between 1.7.0.154 and 8183a606569676b7992a62ad45d2d0a074ca251f:

• Merge branch 'dev'
• update 1.7.1 release notes
• add 1.7.1 release notes
• make the chatInput header a little bit shorter
• fullscreen mode also show tools
• Merge branch 'dev' of github.com:sdcb/chats into dev
• make send button smaller and have 90% opacity
• send button don't taken 1 line in ChatInput
• fix compile error
• don't lock to bottom when generating after any scroll up
• fix: reasoning time is always 0 after refresh
• use ModelProviderIcon for user usage tab
• admin message add message content query
• use pushState for admin message list
• optimize pagination component for 7 pages
• fix a F12 warning for IconReasoning.tsx
• model header ui optimize
• hide upload remote image button for non-vision model
• 优化发送按钮的禁用状态逻辑，并增强 sendMode 的事件监听功能
• fix FE compilation error
• split StepGenerateInfo load ondemand
• optimize generate information
• send button supports mobile
• fix gpt-image-1 issue
• introduce send-button
• fix chat update title issue
• fix canlendar component
• remove en.json
• security log components move to correct component folder
• delete non-used icons
• rename chatIcon to ModelProviderIcon
• ChatIcon support model provider name
• refactor model page icon
• refactor deletePopover
• add isSuccess
• add isSuccess
• refactor to security logs page
• implement tz
• add security audit admin page
• implement audit log/rate limit to sms/password/keycloak login
• ChatInput pause button
• ChatInput position icons in same row