Skip to content

v2.2.3

Latest

Choose a tag to compare

@github-actions github-actions released this 19 Nov 15:07
ec32d83

Downloads (v2.2.3)

Platform Architecture Download
Windows x64 Installer · Portable
Windows arm64 Installer · Portable
macOS Apple silicon (arm64) DMG · ZIP
macOS Intel (x64) DMG · ZIP
Linux x64 AppImage · DEB · RPM
Linux arm64 AppImage · DEB · RPM
Release Notes

Cherry Studio 2.2.3 - Conversation Sorting, New Models, and Agent Reliability

✨ New Features

  • [Conversations] Grouping and sorting are now independent. Sort by last updated, creation date, name (with natural number order), or manual order, and existing custom orders are preserved.
  • [Agent] Import DeepSeek v3 exports into an automatically created DeepSeek Agent, keeping history, branches, reasoning, search records, and available attachments.
  • [Agent] A short next prompt is suggested after replies: press Tab to accept without sending, or Escape to dismiss. Suggestions can be turned off in input settings.
  • [Agent] You can upload an image avatar while creating an Agent.
  • [Agent] Automatic command approval now parses Bash and PowerShell with shared risk rules and asks before each risky command. Approval cards show the requesting task, working directory, and reason. (action required: review permission modes if you rely on the existing prompts; unmatched commands run automatically, while mandatory denials still apply in Full Access.)
  • [Browser] Conversation sidebars support multiple retained tabs with unified styling and context menus, use the configured homepage, and close with their last page.
  • [File Preview] PPTD slide decks now preview in the app with page navigation, zoom, thumbnails, refresh, and a source view.
  • [Files] File paths and Markdown links with line numbers jump to and highlight the referenced line in the preview.
  • [Files] The file selection dialog opens in the folder of the last file you selected during the current session.
  • [Chat] Added an Appearance → Messages preference, "Scroll to the latest message on send" (on by default).
  • [Chat] Long deliverable file lists show the first three files and let you expand or collapse the rest.
  • [Chat] Ctrl + mouse wheel zooms the interface, shows the zoom level while scrolling, and Ctrl+0 resets it.
  • [Knowledge] The URL source accepts several links at once (separated by newlines, commas, or spaces), adds each valid URL separately, and reports invalid entries individually.
  • [Paintings] Generated images show the model and returned size, and are saved in their original format instead of always being converted to PNG.
  • [Computer Use] A new Linux-only "Global input" switch (off by default) lets agents control the shared mouse pointer and system keyboard.
  • [Models] Added GPT-6.2 Astra and Sol, Gemini 4 Carbon and Hydrogen with reasoning controls, Tencent Hunyuan 4 Flash, and Mistral Large 4 reasoning controls.
  • [Models] DeepSeek first-party cost estimates now use weekday peak and off-peak rates while keeping custom prices. (action required: review local cost estimates if you relied on peak-only estimates.)
  • [Models] Enabled providers now appear first in the model service list, and primary model pickers can open in a remembered dialog mode.
  • [Marketplace] Curated skills can be filtered by category, with search and horizontal category navigation.
  • [Composer] Pasting copied folders into Agent conversations adds folder references instead of failing file attachments.
  • [Export] Obsidian exports label each assistant message with the model that produced it.
  • [Shortcuts] Number shortcuts (1–9) have an explicit Off preset in Settings → Shortcuts.
  • [Voice] The recognition model now has an Off option that disables local dictation without removing the saved language or hotkeys.
  • [Code CLI] Qwen Code 1.0 can be installed from its GitHub release.
  • [App] Added Czech (cs-CZ) language support.

🐛 Bug Fixes

  • [Chat] Nested scrollable content such as code previews, editors, and embedded artifacts scrolls independently while the message list scrolls.
  • [Chat] Conversations restored from Settings → Archive open with their messages instead of an empty conversation.
  • [Chat] Interrupted streamed responses are reported as connection errors with partial output kept for retry, instead of an invalid AI response.
  • [Chat] Multi-model fold tabs show every model name again, with a toggle back to compact tabs.
  • [Chat] Chat requests that exceed the model context window recover more reliably.
  • [Notes] LaTeX written with [ … ] and ( … ) renders again in the Notes editor.
  • [Chat] The browser side-panel resize bar is draggable along its full height.
  • [Agent] Assistant text emitted before or between tool calls is visible again in Agent sessions.
  • [Agent] Pi Durable: Stop keeps the session open and the reply paused; reopening after quitting continues an interrupted reply or shows its real result; messages sent while the agent works are saved and answered in order; a withdrawn message is marked "Stopped before sending" with Resend and Edit.
  • [Agent] Messages sent while a turn is running are acknowledged and acted on instead of being silently ignored.
  • [Agent] A pending approval is cleared when its background task ends, so the reply no longer stays on Pending.
  • [Agent] Model selection is scoped to each session, so changing one session no longer changes sibling sessions.
  • [Agent] Claude Code conversations can be forked or edited after tool retries or compaction when the SDK can preserve their context.
  • [Agent] In Accept Edits mode, Claude Code asks before running an escalated Bash command instead of auto-running it by its first word.
  • [Agent] Agent session history repairs message trees during upgrade and prevents edits to in-flight deliveries.
  • [Agent] An Agent turn recovers when a directory occupies its durable store name instead of failing with a network error.
  • [Agent] Reasoning-effort, service-tier, and Fast controls stay available when editing and resending a historical message.
  • [Agent] Historical messages follow the Agent's current avatar, and an avatar chosen late during creation is applied.
  • [Agent] Selection assistant: reused windows no longer open at full conversation size, a closed window no longer leaves a stray capsule, and the panel opens above the active app.
  • [Agent] Agent requests delivered through HTTP or in-process gateways keep protocol metadata, Gemini configuration, generated files, and continuation state.
  • [Agent] Unreadable stored automation tool results are reported to the model as a note instead of breaking the conversation.
  • [Agent] NO_PROXY no longer includes bracketed IPv6 or IPv6 CIDR entries that broke Python HTTP clients in Agent sessions.
  • [Models] Model connection checks now use streaming for streaming-capable models and detect incomplete or failed responses.
  • [Models] Unsupported sampling parameters and penalties are no longer sent to providers, and declared sampling ranges apply after request overrides.
  • [Models] Gemini tool calls no longer fail when a schema contains propertyNames.
  • [Models] A single Cherry Cloud quota refusal no longer locks a model while plan quota remains; the server decides refusals and shows its suggested wait.
  • [Models] CherryIN saved model endpoints refresh automatically from the server while keeping local values when discovery fails.
  • [Models] CherryAI appears first with a paid subscription and last without one.
  • [Models] Provider request rejections without an HTTP status are reported as request errors instead of network interruptions.
  • [Models] Usage stats recover provider cache buckets, and locally computed costs are labeled as estimates.
  • [Models] Agent sessions on OpenAI-compatible, vLLM, or llama.cpp models report how far a request exceeds the context window.
  • [Models] Aborted provider turns now fail instead of reporting a successful stop.
  • [Files] File type icons render offline and stay consistent across file trees, previews, and attachments.
  • [Files] File trees and previews stay current, show save progress and failures, and support scoped refresh without losing local drafts.
  • [Knowledge] The Add URL dialog rejects non-http(s) addresses immediately.
  • [Knowledge] Self-hosted MinerU 4.1 Tasks API parsing is supported while older protocols still work.
  • [Feedback] Problem reports again include logs, activity records, and chat history for the last 24 hours; individual attachment switches were removed.
  • [Scheduled Tasks] Run history recorded before 2.2.0, or whose conversation was deleted, can be opened and shows its stored result.
  • [Skills] Cherry-only built-in skills are no longer exported to the shared ~/.agents/skills directory.
  • [Feishu] Initial channel connections retry with backoff instead of staying stopped after a failed handshake.
  • [UI] Empty assistant, agent, and notes sidebars explain that they are empty instead of showing a blank panel.
  • [Agent] Fixed agent sessions with background subagents slowing down a little more every round.
  • [Agents] Preserve subagent rate-limit error details and show consistent retry status for Pi and Pi Durable tasks.
  • [Agents] Reduce out-of-memory crashes when subagents process many images in parallel.
  • [Cherry AI] Load the model catalog from the server and show sign-in, subscription or AMD authorization requirements; detect delisted models. Built-in Qwen now requires Cherry AI sign-in (action required for users who previously used it signed out).

💄 Improvements

  • [Conversations] Grouping choices are clearer, and list controls are separated from history and management navigation in conversation and Agent session menus.
  • [Agent] The service tier picker explains each tier's trade-offs and shows when a saved tier is unavailable, with provider-specific names in a collapsible details view.
  • [Agent workspace] The folder "Open with" menu now lists every application registered on the system for opening folders, with system icons, instead of only the built-in editors and terminal.
  • [UI] Brand icons, mini-app labels, settings layouts, help, skill controls, and marketplace visuals are more consistent.
  • [UI] Launchpad and Mini Apps icons use consistent sizing and corners with refreshed dark-mode artwork, and inset scrollbars in file trees and lists align with pane edges.
  • [Diagnostics] Native crash reports upload automatically to Sentry when data collection is enabled.
  • [Data upgrade] v1 users must migrate through Cherry Studio 2.2.2; eligible installations offer one-click installation of the migration release, while others keep direct platform-matched downloads. (action required)

⚡ Performance

  • [Agent] Agent sessions with many images or subagents use much less main-process memory and CPU; finished subagents are released and Durable history is read incrementally.
  • [Agent] Restoring large Claude Code sessions and loading their history uses less main-process memory.
  • [Chat] Streaming responses and conversation history loading use fewer copies of conversation snapshots.

What's Changed

  • fix(external-app): list every registered folder application in open with by @DeJeune in #24348
  • fix(release-cos): preserve Windows regional uploads with architecture arguments by @DeJeune in #24446
  • chore: version packages by @github-actions[bot] in #24449
  • fix(release): isolate Windows COS upload dependencies by @DeJeune in #24451
  • chore(release): sync v2.2.2 metadata by @kangfenmao in #24452
  • fix(skills): stop exporting Cherry-only builtins to ~/.agents/skills by @cherry-coding[bot] in #24450
  • docs(chat-context): propose overflow recovery by @DeJeune in #24354
  • fix(chat-context): recover from context overflow by @DeJeune in #24410
  • chore: version packages by @github-actions[bot] in #24456
  • fix(agent): clear pending approval when a background task ends by @cherry-coding[bot] in #24444
  • fix(windows-signing): avoid duplicate SHA-256 signing passes by @DeJeune in #24457
  • perf(release-pipeline): overlap draft uploads and edition packaging by @DeJeune in #24455
  • perf(release-tools): cache and overlap external tool downloads by @DeJeune in #24458
  • perf(packaging): trim duplicate SQLite and Tesseract binaries by @DeJeune in #24462
  • feat(marketplace): filter curated skills by category by @ousugo in #24466
  • fix(scroll-area): align file tree and list scrollbars with pane edges by @DeJeune in #24467
  • feat(session-browser): add retained sidebar browser tabs by @DeJeune in #24468
  • fix(ui): preserve focus for portaled input-group controls by @ousugo in #24474
  • fix(agent): require an acknowledgement for mid-turn steer messages by @cherry-coding[bot] in #24463
  • refactor(data-migration): replace v1 conversion with native installer downloads by @DeJeune in #23998
  • feat(chat-artifacts): collapse long deliverable file lists by @1lck in #24509
  • fix(message-group): restore the fold tab display toggle by @Doekin in #24504
  • fix(agent): keep the avatar file input mounted so a late selection still applies by @cherry-coding[bot] in #24486
  • feat(provider-settings): list enabled providers first in the model service list by @404-Page-Found with @Copilot in #23570
  • feat(agent-composer): suggest the next prompt with Tab acceptance by @1lck in #24506
  • feat(code-cli): install Qwen Code 1.0 from its GitHub release by @angri450 in #24510
  • fix(claude-code): allow SDK-readable history to fork by @WhiteZeek in #24512
  • feat(provider-registry): support Gemini 4 Hydrogen reasoning levels by @DeJeune in #24517
  • feat(models): add Gemini 4 Carbon reasoning effort controls by @cherry-coding[bot] in #24476
  • chore: version packages by @github-actions[bot] in #24519
  • fix(agent-sessions): scope model selection to the session instead of the agent by @404-Page-Found in #20826
  • revert(feedback): restore diagnostic report attachments by @DeJeune in #24532
  • fix(scheduled-tasks): keep stored run results readable without a session by @cherry-coding[bot] in #24465
  • feat(i18n): add Czech (cs-CZ) language by @markiceman in #24494
  • chore: version packages by @github-actions[bot] in #24533
  • fix(snapshot-projection): reduce copies and restrict structuredClone by @DeJeune in #24535
  • test(pi-durable): fix Windows child streaming fixture by @DeJeune in #24544
  • fix(claude-code): bound session archive restore memory by @DeJeune in #24518
  • feat(model-selector): add dialog mode to primary model pickers by @kangfenmao in #24529
  • fix(i18n): translate the model picker mode toggle into Czech by @SiinXu in #24557
  • feat(agent-avatar): upload images during Agent creation by @ousugo in #24530
  • docs(agent-notes): propose pptd slide preview spec by @404-Page-Found in #24488
  • docs(agent-notes): propose service tier explanations spec by @404-Page-Found in #24501
  • feat(provider-billing): add DeepSeek weekday pricing defaults by @WhiteZeek in #24527
  • fix(cherry-cloud): order providers by subscription status by @kangfenmao in #24416
  • fix(i18n): translate the provider billing period strings into Czech by @SiinXu in #24589
  • style(ui): unify settings, brand icons, and marketplace presentation by @kangfenmao in #24574
  • fix(cherryin): refresh saved model endpoints from the server by @kangfenmao in #24531
  • chore: version packages by @github-actions[bot] in #24584
  • fix(sentry): upload native crash dumps automatically by @DeJeune in #24573
  • fix(feishu): retry initial channel connect with backoff by @404-Page-Found in #23376
  • fix(extension-table-plus): point ESM exports and types at the emitted .mjs/.d.mts files by @cherry-coding[bot] in #24590
  • chore: version packages by @github-actions[bot] in #24598
  • fix(api-gateway): fail aborted provider turns instead of a successful stop by @cherry-coding[bot] in #24442
  • fix(conversation-list): clarify grouping and navigation menus by @DeJeune in #24599
  • fix(build): extract Windows tool archives in-process by @cherry-coding[bot] in #24611
  • refactor: remove exports with no remaining consumers by @angri450 in #24600
  • feat(shortcuts): add an explicit Off preset for number shortcuts by @cherry-coding[bot] in #24617
  • fix(error-category): classify status-less invalid_request_error as bad request by @zuhairkhalid229 in #24603
  • test(e2e-regression): wait for the topic sidebar and knowledge import to settle by @cherry-qa[bot] in #24569
  • fix(launchpad): align app icons and dark-mode artwork by @kangfenmao in #24615
  • feat(chat): add a preference for scrolling to the latest message on send by @cherry-coding[bot] in #24597
  • feat(computer-use): honor the user global input preference in SDK actions by @JeremySu0818 in #24595
  • feat(product-guide): add in-app guide skill by @SiinXu in #24542
  • feat(voice): add an Off option to the recognition model to disable dictation by @cherry-coding[bot] in #24566
  • fix(automation-repl): note unreadable history results instead of throwing by @cherry-coding[bot] in #23306
  • fix(agent-identity): send the identity snapshot inside one tag on Pi by @SiinXu in #24621
  • fix(voice-tests): repair main CI preference mock and clone lint by @DeJeune in #24628
  • fix(selection): keep the selection panel above the active app on every show by @cherry-coding[bot] in #24625
  • fix(agent-session): enforce message tree integrity and improve traversal by @DeJeune in #23935
  • fix(composer): paste copied directories as folder tokens by @ZKunZhang in #24629
  • test(e2e): mock embeddings and remove CherryIN model scenarios by @kangfenmao in #24638
  • ci(release): unify Tencent Cloud credentials by @kangfenmao in #24641
  • fix(tool-approval): stop auto-approving acceptEdits Bash by first command word by @kovsu in #24639
  • test(e2e): replace external test providers with local mocks by @kangfenmao in #24645
  • fix(agent-session): read the deleted message once with only the needed columns by @kovsu in #24648
  • fix(tool-schema): stop propertyNames reaching Gemini chat relays by @cherry-coding[bot] in #24664
  • fix(tool-schema): guard the Gemini chat strip against partial scopes by @cherry-coding[bot] in #24667
  • test(tool-schema): provide model in parameterless fixtures by @DeJeune in #24666
  • docs(web-search): propose StepFun text-to-image search spec by @404-Page-Found in #24363
  • docs(agent-notes): propose background system workspace archive by @404-Page-Found in #24361
  • docs(agent-notes): propose cherry-miniapp skill spec by @404-Page-Found in #24359
  • docs(agent-notes): propose session metrics bar spec by @404-Page-Found in #24302
  • fix(provider-settings): log model check timings and document check policy by @404-Page-Found in #24130
  • fix(api-gateway): classify SDK argument failures as request errors by @ZKunZhang in #24640
  • fix(agent): keep historical message avatars current by @ousugo in #24478
  • feat(file-storage): remember last directory in file selection dialog by @404-Page-Found with @Copilot in #23235
  • Keep all assistant text visible in Agent sessions by @404-Page-Found with @Copilot in #23814
  • feat(export): label assistant messages with model name in Obsidian export by @404-Page-Found with @Copilot in #23584
  • fix(agent-composer): keep model controls available while editing by @WhiteZeek in #24694
  • fix(chat-pane): keep the browser resize bar draggable at full height by @DeJeune in #24701
  • fix(heartbeat-editor): stop polling heartbeat status while the dialog is closed by @kovsu in #24680
  • chore(deps): drop adm-zip in favor of archiver, node-stream-zip and jszip by @kovsu in #24679
  • refactor(empty-state): remove the no-op preset prop by @SiinXu in #24683
  • fix(binary-manager): look up system mise without blocking the main process by @kovsu in #24662
  • feat(provider-registry): add new Flash model by @Contentment003111 in #24624
  • chore: version packages by @github-actions[bot] in #24702
  • fix(file): record default-app open handoffs at info level by @cherry-coding[bot] in #24548
  • fix(doctor): load only the incident's preceding messages by @kovsu in #24647
  • fix(e2e-regression): judge a --grep rerun by the selection its own run recorded by @cherry-qa[bot] in #24681
  • fix(notes): restore [...] and (...) math in the Notes editor by @cherry-coding[bot] in #24576
  • fix(agent-session): recover a directory occupying the durable store name by @cherry-coding[bot] in #24374
  • fix(usage): recover provider cache buckets and label computed costs as estimates by @cherry-coding[bot] in #24715
  • fix(model-check): probe streaming-capable models with complete streams by @WhiteZeek in #24724
  • feat(command-policy): parse shell invocations and apply JSON approval rules by @1lck in #24434
  • feat(error-reporting): classify failures before Sentry capture by @DeJeune in #21720
  • fix(cherry-cloud): let the server decide quota refusals instead of the client by @SiinXu in #24687
  • docs(agent-prediction): specify native runtime suggestions by @DeJeune in #24717
  • feat(agent-runtime): use native prompt prediction for Claude and Pi by @DeJeune in #24722
  • fix(empty-state): explain empty assistant, agent and note sidebars by @SiinXu in #24684
  • fix(open-mineru): support MinerU 4.1 Tasks parsing by @DeJeune in #24745
  • feat(provider-registry): support GPT-6.2 Astra and Sol by @DeJeune in #24749
  • chore: version packages by @github-actions[bot] in #24752
  • feat(zoom): support Ctrl + mouse wheel to zoom the interface by @cherry-coding[bot] in #24730
  • docs(file-preview): specify file link line navigation by @DeJeune in #24746
  • chore(lint-tooling): adopt tsrs for lint and type checks by @DeJeune in #24747
  • feat(file-preview): navigate file links to source lines by @DeJeune in #24750
  • fix(window-manager): clear per-window state when pooled windows return to the pool by @kovsu in #24162
  • fix(changeset-policy): reject ineligible and unchanged packages by @DeJeune in #24773
  • fix(file-observation): keep file trees and previews consistent by @DeJeune in #24728
  • fix(model-parameters): enforce active model sampling limits by @DeJeune in #24776
  • chore: version packages by @github-actions[bot] in #24785
  • fix(native-rebuild): drop the obsolete rebuild:node step by @DeJeune in #24789
  • chore(lint): enforce import layout with oxlint by @DeJeune in #24790
  • feat(provider-registry): expose Mistral Large 4 reasoning controls by @404-Page-Found with @Copilot in #23236
  • feat(file-preview): PPTD slide preview plugin by @404-Page-Found in #24796
  • chore: version packages by @github-actions[bot] in #24800
  • fix(agent-proxy): keep unparseable IPv6 entries out of the Agent NO_PROXY by @cherry-coding[bot] in #24793
  • feat(model-speed-control): explain service tiers and show saved vs effective selection by @404-Page-Found in #24788
  • chore(deps): drop redundant @types/uuid by @kovsu in #24810
  • fix(knowledge): validate source URLs with the shared http(s) check by @kovsu in #24812
  • fix(selection-agent): stop a stale session from resizing a reused window by @kovsu in #24811
  • feat(knowledge): accept multiple URLs in one add-dialog submission by @cherry-coding[bot] in #24794
  • feat(paintings): show the model and returned size of a generated image, and save the original file by @cherry-coding[bot] in #24807
  • fix(api-gateway): report overflow token counts in Claude Code's format by @kovsu in #24830
  • chore: version packages by @github-actions[bot] in #24832
  • fix(agent-session): bound trace and native observation memory by @kangfenmao in #24500
  • docs(agent-notes): propose agent image byte paths spec by @DeJeune in #24650
  • perf(agent-runtime): share frozen observation values instead of copying history by @DeJeune in #24651
  • perf(file-processor): validate each native history image once by @DeJeune in #24661
  • perf(agent-session): release finished Durable subagents by @DeJeune in #24705
  • perf(agent-runtime-pi-durable): read only new Durable history on repeated reads by @DeJeune in #24714
  • docs(agent-notes): propose deriving Durable facts and observations as folds by @DeJeune in #24813
  • perf(agent-runtime-pi-durable): derive behavior facts from in-memory folds by @DeJeune in #24814
  • perf(agent-runtime-pi-durable): publish observation deltas from a history fold by @DeJeune in #24817
  • perf(agent-session): apply Durable deltas without per-update rescans by @DeJeune in #24818
  • docs(pi-durable): propose Durable turn lifecycle authority by @DeJeune in #24759
  • fix(pi-durable): report a Durable abort as a paused turn by @DeJeune in #24760
  • fix(pi-durable): keep the Durable access open on Stop by @DeJeune in #24766
  • fix(pi-durable): recover interrupted replies from the Durable record by @DeJeune in #24774
  • fix(pi-durable): admit busy-time input in the Durable queue by @DeJeune in #24777
  • test(pi-durable): verify turn authority in Electron and mark the Spec implemented by @DeJeune in #24781
  • docs(agent-import): specify DeepSeek import into Agent sessions by @DeJeune in #24729
  • refactor(agent-runtime): replace internal HTTP codecs with protocol objects by @DeJeune in #24786
  • test(agent-session): fix historical tree migration fixtures by @DeJeune in #24851
  • feat(agent-import): import DeepSeek history into Agent sessions by @DeJeune in #24738
  • docs(conversation-list): specify independent sorting by @DeJeune in #24836
  • feat(conversation-list): separate grouping and sorting by @DeJeune in #24838
  • fix(topics): keep restored conversations openable from the sidebar by @cherry-coding[bot] in #24852
  • fix(chat): keep nested scrollers hittable while the message list scrolls by @cherry-coding[bot] in #24848
  • fix(release): backport #24853 to v2.2.3 by @kangfenmao in #24855
  • fix(release): backport #24859 to v2.2.3 by @DeJeune in #24860
  • fix(release): backport #24876 to v2.2.3 by @kangfenmao in #24877
  • test(pi-durable): backport overflow budget to release v2.2.3 by @DeJeune in #24880
  • fix(agent-session): backport #24913 to v2.2.3 by @DeJeune in #24935
  • fix(release): backport #24919 to v2.2.3 by @kangfenmao in #24941
  • fix(cherry-cloud): backport #24731 to v2.2.3 by @DeJeune in #24943
  • test(pi-durable): backport #24945 to v2.2.3 by @DeJeune in #24946

New Contributors

Full Changelog: v2.2.2...v2.2.3