AIRI

AIMIT

AIRI release notes.

Latest v0.11.3 · by AIRIWritten in TypeScriptWebsitemoeru-ai/airiRSS

Release activity

Release activity — 14 releases across 13 days since Apr 6, 2026. Each cell is one day; darker means more releases that day. Nothing is recorded before Apr 6, 2026. Older weeks are hidden at this screen width.
JunJulAugSep
SundayNo releases on May 24, 2026No releases on May 31, 2026No releases on Jun 7, 2026No releases on Jun 14, 2026No releases on Jun 21, 2026No releases on Jun 28, 2026No releases on Jul 5, 2026No releases on Jul 12, 2026No releases on Jul 19, 2026No releases on Jul 26, 2026No releases on Aug 2, 2026No releases on Aug 9, 2026No releases on Aug 16, 2026No releases on Aug 23, 2026No releases on Aug 30, 2026No releases on Sep 6, 2026
MondayNo releases on May 25, 2026No releases on Jun 1, 2026No releases on Jun 8, 2026No releases on Jun 15, 2026No releases on Jun 22, 2026No releases on Jun 29, 2026No releases on Jul 6, 2026No releases on Jul 13, 2026No releases on Jul 20, 2026No releases on Jul 27, 2026No releases on Aug 3, 2026No releases on Aug 10, 2026No releases on Aug 17, 20261 release on Aug 24, 2026No releases on Aug 31, 2026No releases on Sep 7, 2026
TuesdayNo releases on May 26, 2026No releases on Jun 2, 2026No releases on Jun 9, 2026No releases on Jun 16, 2026No releases on Jun 23, 2026No releases on Jun 30, 2026No releases on Jul 7, 2026No releases on Jul 14, 2026No releases on Jul 21, 2026No releases on Jul 28, 2026No releases on Aug 4, 2026No releases on Aug 11, 2026No releases on Aug 18, 2026No releases on Aug 25, 2026No releases on Sep 1, 2026No releases on Sep 8, 2026
WednesdayNo releases on May 27, 2026No releases on Jun 3, 2026No releases on Jun 10, 2026No releases on Jun 17, 2026No releases on Jun 24, 2026No releases on Jul 1, 20261 release on Jul 8, 2026No releases on Jul 15, 2026No releases on Jul 22, 2026No releases on Jul 29, 2026No releases on Aug 5, 2026No releases on Aug 12, 2026No releases on Aug 19, 2026No releases on Aug 26, 2026No releases on Sep 2, 2026No releases on Sep 9, 2026
ThursdayNo releases on May 28, 2026No releases on Jun 4, 2026No releases on Jun 11, 2026No releases on Jun 18, 2026No releases on Jun 25, 2026No releases on Jul 2, 2026No releases on Jul 9, 2026No releases on Jul 16, 2026No releases on Jul 23, 2026No releases on Jul 30, 2026No releases on Aug 6, 2026No releases on Aug 13, 2026No releases on Aug 20, 20262 releases on Aug 27, 2026No releases on Sep 3, 2026No releases on Sep 10, 2026
FridayNo releases on May 29, 2026No releases on Jun 5, 2026No releases on Jun 12, 2026No releases on Jun 19, 2026No releases on Jun 26, 2026No releases on Jul 3, 2026No releases on Jul 10, 20261 release on Jul 17, 2026No releases on Jul 24, 2026No releases on Jul 31, 2026No releases on Aug 7, 2026No releases on Aug 14, 2026No releases on Aug 21, 2026No releases on Aug 28, 2026No releases on Sep 4, 2026
SaturdayNo releases on May 30, 2026No releases on Jun 6, 2026No releases on Jun 13, 2026No releases on Jun 20, 2026No releases on Jun 27, 2026No releases on Jul 4, 2026No releases on Jul 11, 20261 release on Jul 18, 2026No releases on Jul 25, 2026No releases on Aug 1, 2026No releases on Aug 8, 2026No releases on Aug 15, 2026No releases on Aug 22, 20261 release on Aug 29, 2026No releases on Sep 5, 2026

14 releases since Apr 6, 2026, busiest day 2

Changelog

v0.12.0-beta.5

Pre-release
Added 6
  • Select the MAGIC driver to generate Neuro-sama like motion (aka Neuro bump) in Live2D
  • MAGIC driver includes bundled Idle/calm and Speaking/excited profiles with speech lip sync preservation and forward-facing character look
  • Live2D imports now show a clear report listing model data, absent references, unreferenced resources, and steps to fix each issue
  • Use VOICEVOX Engine or AivisSpeech Engine as a local speech provider with connection and voice controls
  • New internal gamepad packages normalize browser controllers, provide Vue lifecycle integration, and add PlayStation 5 DualSense USB and Bluetooth support
  • Live2D Motion Control workbench now supports recording, editing, playback, filtering, keyboard and controller input, and procedural motion generation
Changed 2
  • @proj-airi/server-runtime now gives each port to one listener and disables reusePort to prevent unsupported socket errors
  • Fresh Electron 42 setups now install Electron before dev or start to prevent the desktop app from failing before launch
Fixed 7
  • Live2D ZIP archives created on macOS now import correctly with hidden metadata no longer preventing model file and expression discovery
  • Chat now reliably creates or restores a session during startup in single-window and multi-window setups
  • Short mobile chat histories now sit above the composer with new messages staying near the bottom
  • Chat, settings, onboarding, and other panels now use consistent scroll areas with the chat composer remaining fixed on desktop
  • On iOS, microphone access now opens the system permission prompt instead of closing the app
  • The desktop local server now starts reliably on Windows and macOS with socket setup no longer requesting unsupported port reuse
  • The desktop title bar no longer shows an inactive information icon
Removed 1
  • The retired transcription package and its stale CI command are gone

From AIRI

v0.12.0-beta.5 Highlights
New Contributors
Product updates
Live2D
  • You can now select the MAGIC1 driver to generate Neuro-sama like motion (aka Neuro bump).
  • The driver includes bundled Idle / calm and Speaking / excited profiles. It can preserve speech lip sync and keep the character looking forward.
  • Live2D imports now show a clear report before import. It lists model data, absent references, unreferenced resources, and steps to fix each issue.
  • Live2D ZIP archives created on macOS now import correctly. Hidden metadata no longer prevents the importer from finding model files and expressions.
Speech
  • You can now use VOICEVOX Engine or AivisSpeech Engine as a local speech provider. Each provider includes connection and voice controls.
Bug fixes and polish
  • Chat now reliably creates or restores a session during startup in single-window and multi-window setups.
  • Short mobile chat histories now sit above the composer. New messages stay near the bottom, while upward scrolling keeps your reading position.
  • Chat, settings, onboarding, and other panels now use consistent scroll areas. On desktop, the chat composer remains fixed while the history scrolls.
  • On iOS, microphone access now opens the system permission prompt instead of closing the app. The app remains open after you deny access.
  • The desktop local server now starts reliably on Windows and macOS. Socket setup no longer requests unsupported port reuse.
  • The desktop title bar no longer shows an inactive information icon.
To developers
  • @proj-airi/server-runtime now gives each port to one listener and disables reusePort. This prevents unsupported socket errors across desktop platforms.
To contributors
  • New internal gamepad packages normalize browser controllers, provide Vue lifecycle integration, and add PlayStation 5 DualSense USB and Bluetooth support.
  • The Live2D Motion Control workbench now supports recording, editing, playback, filtering, keyboard and controller input, and procedural motion generation.
  • Fresh Electron 42 setups now install Electron before dev or start, so the desktop app no longer fails before launch.
  • The retired transcription package and its stale CI command are gone. Nix dependency hashes and the documented typecheck command are now current.

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.12.0-beta.4...v0.12.0-beta.5

Footnotes
  1. MAGIC stands for Markovian Animation Generator with Illusory Conditioning

View originalPermalink
How v0.12.0-beta.5 went

v0.12.0-beta.4

Pre-release
Fixed 2
  • Live2D avatars now remain smooth during long sessions with reduced pauses from heavy memory use
  • On mobile Safari, tapping the message input while the keyboard closes now restores the keyboard correctly

From AIRI

v0.12.0-beta.4 Highlights
Bug fixes
  • Live2D avatars now remain smooth during long sessions, with fewer pauses caused by heavy memory use.
  • On mobile Safari, tapping the message input while the keyboard closes now restores the keyboard correctly.

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.12.0-beta.3...v0.12.0-beta.4

View originalPermalink
How v0.12.0-beta.4 went

v0.12.0-beta.3

Pre-release
Added 3
  • Apple Speech is now available for on-device transcription on macOS 26 or later without requiring an API key, with locale selection support
  • Long-press and drag the message input upward on mobile to dock it as a compact keyboard bubble, and tap to restore it
  • Provider definitions can now load settings schemas, onboarding fields, validators, availability, and runtime instances asynchronously
Changed 2
  • The hosted authentication UI no longer bundles the Stage renderer, model assets, or local inference runtime
  • Workspace installs now build FFmpeg binaries for macOS x64 and Linux ARM64
Fixed 8
  • Sign-in now opens in the system browser on iOS and Android and returns you to AIRI
  • AIRI keeps its main window, captions, and widgets visible more reliably on Linux with Wayland
  • AIRI now respects explicit X11 and Wayland choices and avoids unsupported media-permission calls on Linux
  • Safari no longer loses the software keyboard when tapping the message input while the keyboard closes
  • OpenAI and OpenAI-compatible provider checks now support newer completion-token rules, allowing more models to complete setup successfully
  • Provider choices now match your sign-in state, with account-backed providers disappearing after sign-out while local and browser providers remain available
  • Connection status indicators now use a steady warning animation when AIRI is offline
  • Provider settings now show loading and retry states when AIRI discovers runtime options

From AIRI

v0.12.0-beta.3 Highlights

This cumulative update covers changes since v0.12.0-beta.1.

New Contributors
New features
Desktop ver.
  • You can now use Apple Speech for on-device transcription on macOS 26 or later. It needs no API key and includes locale selection.
Mobile ver.
  • You can now long-press the message input and drag it upward to dock it as a compact keyboard bubble. Tap it to restore the input.
Important fixes
  • Sign-in now opens in the system browser on iOS and Android, then returns you to AIRI.
  • AIRI now keeps its main window, captions, and widgets visible more reliably on Linux with Wayland.
  • AIRI now respects explicit X11 and Wayland choices and avoids unsupported media-permission calls on Linux.
  • Safari no longer loses the software keyboard when you tap the message input while the keyboard closes.
  • OpenAI and OpenAI-compatible provider checks now support newer completion-token rules. As a result, more models complete setup successfully.
  • Provider choices now match your sign-in state. Account-backed providers disappear after sign-out, while local and browser providers stay available.
  • Connection status indicators now use a steady warning animation when AIRI is offline.
  • Provider settings now show loading and retry states when AIRI discovers runtime options.
To developers
  • Provider definitions can now load settings schemas, onboarding fields, validators, availability, and runtime instances asynchronously.
  • The hosted authentication UI no longer bundles the Stage renderer, model assets, or local inference runtime.
To contributors
  • The architecture diagram now reflects the current applications, shared packages, integrations, desktop server channel, and hosted backend.
  • The guides now explain upstream setup, secure Base URLs, LAN use with LM Studio, and CometAPI model IDs.
  • The Korean documentation now uses corrected spacing and current setup commands.
  • The shared Polaroid devtool now lives in stage-pages. Stage UI imports now use public package exports.
  • The repository refreshed its dependencies and Nix dependency hash.
  • CI now uses pnpm/setup with Node 26 and Wrangler v4. Vitest jobs now use stable local-storage behavior.
  • Workspace installs now build FFmpeg binaries for macOS x64 and Linux ARM64.
Upgrade notes
  • Contributors must use Node.js 26.7.0 and pnpm 11.24.0. The repository pins these versions in .tool-versions and packageManager.

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.12.0-beta.1...v0.12.0-beta.3

View originalPermalink
How v0.12.0-beta.3 went

v0.12.0-beta.1

Pre-release
Added 10
  • Desktop version now supports MMD models with motion import and adjustment of physics, gaze, lip sync, expressions, and materials
  • Desktop version now supports Tachie archives that switch complete character images to match emotions
  • VRM characters now react when you click their heads, arms, hands, or feet
  • New Hearing playground lets you monitor the active speech-recognition pipeline and review each completed speech segment
  • Desktop version can relay conversations to a connected Minecraft bot and read supported bot speech aloud
  • Mobile version now has a permission panel and clearer onboarding
Changed 18
  • Live2D imports now support CJK paths and older filename encodings
  • You can now mute voice output across sessions or stop all current speech from the Controls Island
  • Speech recognition now keeps transcript corrections and handles voice activity, pauses, and streaming audio more reliably
  • Long conversations use less rendering work so chat remains responsive as the history grows
  • Tool messages now remain available across chat turns
  • Streamed messages no longer flash and tool-call progress stays synchronized
Fixed 8
  • A provider settings loop could leave AIRI loading forever; the startup and synchronization path is now stable
  • Desktop chat creation, switching, deletion, and import refresh now work reliably
  • Chat and sign-in state now remain consistent across Electron windows
  • AIRI Card drafts, active sessions, avatar Models, and runtime data now remain intact during edits and activation
  • Whisper worker errors now stop promptly instead of leaving transcription stuck
  • Stopping Kokoro speech no longer destroys the worker needed for the next response
  • Safari no longer blocks the authentication interface while it loads code chunks
  • If a Live2D model fails to render, the Stage now shows a clear error and remains usable

From AIRI

v0.12.0-beta.1 Highlights

This beta adds new character formats, stronger desktop controls, and major reliability work across chat, speech recognition, and provider settings.

New Contributors
Product updates
Models and Stage
  • Desktop ver. now supports MMD models. You can import motions and adjust physics, gaze, lip sync, expressions, and materials.
  • Desktop ver. now supports Tachie archives that switch complete character images to match emotions.
  • VRM characters now react when you click their heads, arms, hands, or feet.
  • Live2D imports now support CJK paths and older filename encodings.
  • If a Live2D model fails to render, the Stage now shows a clear error and remains usable.
Chat, Hearing, and Speech
  • You can now mute voice output across sessions or stop all current speech from the Controls Island.
  • The new Hearing playground lets you monitor the active speech-recognition pipeline and review each completed speech segment.
  • Speech recognition now keeps transcript corrections and handles voice activity, pauses, and streaming audio more reliably.
  • Long conversations use less rendering work, so chat remains responsive as the history grows.
  • Tool messages now remain available across chat turns.
  • Streamed messages no longer flash, and tool-call progress stays synchronized.
  • Chat math now uses consistent syntax for predictable display.
Desktop and Mobile
  • The Controls Island now moves to the nearest window corner after you reposition the desktop window.
  • Connection status now lives inside the Controls Island, and tray-based window moves use animations.
  • Desktop pairing settings and QR codes now stay synchronized.
  • Desktop ver. can relay conversations to a connected Minecraft bot and read supported bot speech aloud.
  • AIRI now starts on Linux ARM64. Flatpak installs also include desktop and AppStream metadata.
  • Mobile ver. now has a permission panel, clearer onboarding, and Android microphone selection.
  • On mobile browsers, the software keyboard no longer hides or shifts the chat input unexpectedly.
Accounts and providers
  • You can enable Thinking for supported Consciousness models from one model option.
  • DeepSeek now applies its reasoning mode correctly.
  • Provider validation status now persists, and each provider uses its own validation rules.
  • Tool use works with more providers because AIRI adapts tool schemas to each provider.
  • Volcengine Coding Plan now lists current models with localized descriptions.
  • Steam sign-in and account linking are now available.
  • Hosted authentication now supports Apple sign-in and native Apple ID tokens.
  • Account deletion now revokes linked social authorizations.
  • Hosted model routing can fall back across configured provider groups when one route is unavailable.
Important fixes
  • A provider settings loop could leave AIRI loading forever. The startup and synchronization path is now stable.
  • Desktop chat creation, switching, deletion, and import refresh now work reliably.
  • Chat and sign-in state now remain consistent across Electron windows.
  • AIRI Card drafts, active sessions, avatar Models, and runtime data now remain intact during edits and activation.
  • Whisper worker errors now stop promptly instead of leaving transcription stuck.
  • Stopping Kokoro speech no longer destroys the worker needed for the next response.
  • Safari no longer blocks the authentication interface while it loads code chunks.
To developers
  • The CCv3 codec now accepts older and newer cards while preserving unknown fields for forward compatibility.
  • Hosted chat WebSockets now expose versioned Eventa contracts. Version 2 authenticates after connection and enforces payload limits.
  • The desktop plugin SDK now preserves nullable enum schemas correctly.
  • MMD and Tachie renderer packages expose reusable scene components and runtime controls.
  • Hosted configuration values can reload from PostgreSQL without a service restart.
To contributors
  • Audio tests can now drive the full input and output pipeline with deterministic microphone fixtures.
  • Vitest browser projects now run headlessly in CI.
  • Vishot now lives in its public repository. AIRI also has updated capture and pull-request skills.
  • Polaroid captures now include imported Live2D models.
  • The new i18n glossary and Crowdin workflow keep product terms consistent across translations.
  • Release workflows now support Steam desktop deployment, image provenance, SBOM attestations, and automated README artifact updates.
  • The documentation site now includes Korean content and smoother search, theme, and page transitions.
Upgrade notes
  • Run the Resource API and Auth as separate services.
  • Set AUTH_SERVER_URL, AUTH_SERVER_INTERNAL_URL, PUBLIC_URL, and RESOURCE_SERVER_URL for the new service boundary.
  • On Railway, set RATE_LIMIT_TRUSTED_PROXY=railway only for services that receive Railway proxy traffic.
  • Do not run shared database migrations from Auth. The Resource API remains the only migration owner.
  • Before you build or run Godot tests, install the .NET 10 SDK.

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.11.3...v0.12.0-beta.1

View originalPermalink
How v0.12.0-beta.1 went

v0.11.3

Latest
Added 4
  • Voice input is now available directly in the desktop companion, with microphone buffering and echo avoidance during speech
  • Desktop window can now be moved back to the center of the current display from the controls island or desktop data settings
  • Character cards can now be imported and exported as AIRI .zip packages
  • Exported character packages include the character card plus supported display model resources
Changed 7
  • Microphone permission handling is more reliable in the Electron app while limiting permission grants to AIRI-owned local pages
  • Desktop sign-in is more robust with improved auth callback handling and support for newer AIRI backend origin
  • Settings entry icons are now preserved in production builds
  • Live2D model loading now handles non-ASCII file names inside restored model folders and avoids double-encoding already encoded paths
  • AIRI now records AI generation, LLM usage, Stripe identity, and checkout-related analytics more consistently
  • PostHog event naming and forwarding were normalized across app and server surfaces
  • Auth callback handling now validates OIDC state inside the loopback server before resolving the login attempt
Fixed 1
  • Invalid or incomplete card packages now show clearer import errors
Removed 1
  • Character card import no longer accepts raw .json card files and now expects AIRI .zip packages instead

From AIRI

v0.11.3 Highlights
New Contributors
Product Updates
Desktop Companion
  • You can now use voice input directly in the desktop companion. AIRI listens through the configured microphone, buffers short transcript fragments, and avoids feeding its own speech back into chat while it is talking.
  • You can now move the desktop window back to the center of the current display from the controls island or the desktop data settings. This helps recover AIRI when the window gets awkwardly positioned or ends up hard to reach.
  • Microphone permission handling is more reliable in the Electron app, while still limiting permission grants to AIRI-owned local pages.
  • Desktop sign-in is more robust. Invalid auth callbacks no longer consume the one-time login flow, and the standalone auth UI now accepts the newer AIRI backend origin.
  • Settings entry icons are now preserved in production builds, so settings navigation should no longer lose icons after bundling.
Character Cards And Models
  • Character cards can now be imported and exported as AIRI .zip packages.
  • Exported packages include the character card plus supported display model resources, so sharing a character can carry more of the setup with it.
  • Invalid or incomplete card packages now show clearer import errors.
  • Live2D model loading now handles non-ASCII file names inside restored model folders and avoids double-encoding already encoded paths.
Account And Cloud Reliability
  • AIRI now records AI generation, LLM usage, Stripe identity, and checkout-related analytics more consistently. This improves billing, support investigation, and product health tracking without changing the visible checkout flow.
  • PostHog event naming and forwarding were normalized across app and server surfaces, making account and usage events easier to correlate.
To Developers
  • @proj-airi/pipelines-audio received stronger event typing, playback manager cleanup, ACT payload normalization, transcript buffering, and clearer error handling for speech and playback flows.
  • Chat/LLM runtime paths now carry usage metadata more consistently, including server-side OpenAI-compatible request handling and downstream analytics headers.
  • Auth callback handling now validates OIDC state inside the loopback server before resolving the login attempt.
To Contributors
  • Grafana server dashboards and related build tests were refreshed.
  • .env files are now ignored to reduce accidental credential leakage.
  • Nix Xorg attributes were updated for current package structure.
  • Claude Code dependencies, sponsor data, and generated translations were refreshed.
  • The release bumps AIRI packages and apps to 0.11.1.
Upgrade Notes
  • Character card import now expects AIRI .zip packages instead of raw .json card files.

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.11.0...v0.11.1

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.11.0...v0.11.3

View originalPermalink
How v0.11.3 went

v0.11.1

Pre-release
Added 4
  • Voice input is now available directly in the desktop companion, listening through the configured microphone and buffering short transcript fragments while avoiding feedback from AIRI's own speech
  • Desktop window can now be moved back to the center of the current display from the controls island or desktop data settings
  • Character cards can now be imported and exported as AIRI .zip packages
  • Exported character packages include the character card plus supported display model resources
Changed 10
  • Microphone permission handling in the Electron app is more reliable while limiting permission grants to AIRI-owned local pages
  • Desktop sign-in is more robust, with invalid auth callbacks no longer consuming the one-time login flow and standalone auth UI accepting newer AIRI backend origin
  • Settings entry icons are now preserved in production builds
  • Character card import now expects AIRI .zip packages instead of raw .json card files
  • Live2D model loading now handles non-ASCII file names inside restored model folders and avoids double-encoding already encoded paths
  • AIRI now records AI generation, LLM usage, Stripe identity, and checkout-related analytics more consistently
Fixed 1
  • Invalid or incomplete card packages now show clearer import errors

From AIRI

v0.11.1 Highlights
New Contributors
Product Updates
Desktop Companion
  • You can now use voice input directly in the desktop companion. AIRI listens through the configured microphone, buffers short transcript fragments, and avoids feeding its own speech back into chat while it is talking.
  • You can now move the desktop window back to the center of the current display from the controls island or the desktop data settings. This helps recover AIRI when the window gets awkwardly positioned or ends up hard to reach.
  • Microphone permission handling is more reliable in the Electron app, while still limiting permission grants to AIRI-owned local pages.
  • Desktop sign-in is more robust. Invalid auth callbacks no longer consume the one-time login flow, and the standalone auth UI now accepts the newer AIRI backend origin.
  • Settings entry icons are now preserved in production builds, so settings navigation should no longer lose icons after bundling.
Character Cards And Models
  • Character cards can now be imported and exported as AIRI .zip packages.
  • Exported packages include the character card plus supported display model resources, so sharing a character can carry more of the setup with it.
  • Invalid or incomplete card packages now show clearer import errors.
  • Live2D model loading now handles non-ASCII file names inside restored model folders and avoids double-encoding already encoded paths.
Account And Cloud Reliability
  • AIRI now records AI generation, LLM usage, Stripe identity, and checkout-related analytics more consistently. This improves billing, support investigation, and product health tracking without changing the visible checkout flow.
  • PostHog event naming and forwarding were normalized across app and server surfaces, making account and usage events easier to correlate.
To Developers
  • @proj-airi/pipelines-audio received stronger event typing, playback manager cleanup, ACT payload normalization, transcript buffering, and clearer error handling for speech and playback flows.
  • Chat/LLM runtime paths now carry usage metadata more consistently, including server-side OpenAI-compatible request handling and downstream analytics headers.
  • Auth callback handling now validates OIDC state inside the loopback server before resolving the login attempt.
To Contributors
  • Grafana server dashboards and related build tests were refreshed.
  • .env files are now ignored to reduce accidental credential leakage.
  • Nix Xorg attributes were updated for current package structure.
  • Claude Code dependencies, sponsor data, and generated translations were refreshed.
  • The release bumps AIRI packages and apps to 0.11.1.
Upgrade Notes
  • Character card import now expects AIRI .zip packages instead of raw .json card files.

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.11.0...v0.11.1

View originalPermalink
How v0.11.1 went

v0.11.0

Added 11
  • You can now stop AIRI's speech manually from the stage UI, including the desktop and mobile layouts
  • Tool calls can now be retried from chat
  • Google Gemini TTS is now available as a speech provider
  • Aliyun NLS ASR, StepFun TTS, default TTS model handling, generated voice previews, and official provider catalog routing are now part of AIRI's official speech stack
  • Vision provider settings were added for providers such as Ollama, LM Studio, Cloudflare Workers AI, Azure AI Foundry, Amazon Bedrock, and official providers
  • Spine 2D model support is now integrated, with model settings, preview support, animation handling, and runtime validation
Changed 8
  • Chat sync is more reliable when moving between desktop stage routes
  • Speech settings now support Voice Pack model selection, recommended voices, and better handling for no speech setups
  • Voice input and transcription flows were cleaned up, including hearing controls, audio device handling, and streaming transcription behavior
  • Chat actions now give clearer copy feedback, and desktop connection settings now expose copy-friendly connection details
  • Live2D and VRM behavior is smoother: lip sync is less likely to be overridden by motions, Live2D zip loading is more reliable, and several missing default values were restored
  • Experimental Godot stage sidecar gained VRM runtime import, camera control, a new visual baseline, avatar glow, and rim lighting
  • The account UI moved to accounts.airi.build
  • Plugin terminology and APIs moved toward extension naming, with new extension entrypoint/session APIs and updated SDK surfaces
Fixed 1
  • Language choice now persists correctly in the Electron app

From AIRI

v0.11.0 Highlights
Product updates
Chat, voice, and stage
  • You can now stop AIRI's speech manually from the stage UI, including the desktop and mobile layouts.
  • Tool calls can now be retried from chat, so failed extension/gamelet actions do not always require starting over.
  • Chat sync is more reliable when moving between desktop stage routes.
  • Speech settings now support Voice Pack model selection, recommended voices, and better handling for "no speech" setups.
  • Voice input and transcription flows were cleaned up, including hearing controls, audio device handling, and streaming transcription behavior.
  • Chat actions now give clearer copy feedback, and desktop connection settings now expose copy-friendly connection details.
New providers and models
  • Google Gemini TTS is now available as a speech provider.
  • AIRI's official speech stack now includes Aliyun NLS ASR, StepFun TTS, default TTS model handling, generated voice previews, and official provider catalog routing.
  • Vision provider settings were added for providers such as Ollama, LM Studio, Cloudflare Workers AI, Azure AI Foundry, Amazon Bedrock, and official providers.
Avatars and stage rendering
  • Spine 2D model support is now integrated, with model settings, preview support, animation handling, and runtime validation.
  • Live2D and VRM behavior is smoother: lip sync is less likely to be overridden by motions, Live2D zip loading is more reliable, and several missing default values were restored.
  • The experimental Godot stage sidecar gained VRM runtime import, camera control, a new visual baseline, avatar glow, and rim lighting.
Desktop fixes
  • Desktop now prevents duplicate app instances.
  • Language choice persists correctly in the Electron app.
  • Tray resizing now anchors to the dominant display.
  • The updater path and exit behavior were fixed.
  • Several widget, Spine preview, tooltip, Unicode, and hearing-control issues were cleaned up.
Account and billing
  • Account linking now works better across different email providers, with clearer OAuth account selection and error handling.
  • The account UI moved to accounts.airi.build.
  • FLUX checkout from the Electron desktop renderer was unblocked.
  • Billing for server-side usage is stricter and more reliable, including partial-debit handling to prevent unpaid usage gaps.
Fixes worth scanning for
  • Provider validation now only applies to providers that are actually listed.
  • Provider source filters now use corrected categories.
  • Artistry now defaults to no provider instead of selecting one unexpectedly.
  • Character card module selections now sync back to the active character card.
  • Streaming TTS requests are gated more safely, and TTS session isolation was hardened.
  • Official voice preview URLs are now exposed correctly.
  • Auth callback trust handling was tightened.
  • Social-only email sign-in hints are clearer.
To developers
  • The server now has an in-process LLM/TTS gateway with official alias routing, upstream attempt propagation, structured OpenAI-compatible request logging, and streaming TTS catalog support.
  • Server health checks moved to Kubernetes-style /livez and /readyz probes.
  • The official provider catalog is now managed server-side, including catalog gating and official ASR/LLM alias policy.
  • better-ws was added and server runtime WebSocket handling was refactored around liveness, routing, codecs, and response handling.
  • Plugin terminology and APIs moved toward "extension" naming, with new extension entrypoint/session APIs and updated SDK surfaces.
  • Gamelet/widget integrations now support coordinated iframe requests and retriable tool calls.
  • The DashScope CosyVoice adapter moved to the v2 two-step REST flow.
  • Docker builds now install needed build dependencies and run package build steps more reliably.
To contributors
  • cap-vite now auto-selects the first available native device when no target is specified.
  • Playwright was upgraded to avoid install hangs on Node 24.
  • CI now repairs pnpm lockfile snapshots, automates sponsor-update PRs, mirrors release assets to GitCode, and includes Godot stage publishing.
  • The workspace dependency setup moved further into catalogs.
  • Docs were refreshed with desktop quick-start updates, install tips, AtomGit acknowledgements, and a character card template reference.
  • Computer-use MCP gained terminal screen heuristics, bounded terminal capture, background scheduler improvements, browser repair suggestions, and safer cleanup for timed-out terminal processes.
  • The Minecraft service gained stronger perception, reflex, combat, and planning behavior.
Upgrade notes
  • Server operators should update health checks from legacy endpoints to /livez and /readyz.
  • Deployments using the old auth UI host should move to accounts.airi.build.
  • Server TTS deployments should review the newer UNSPEECH_UPSTREAM configuration and official provider catalog setup.
  • Extension/plugin authors should review the plugin-to-extension rename and update manifests/APIs where applicable.
View originalPermalink
How v0.11.0 went

v0.10.2

Added 3
  • Use Xiaomi MiMo for speech and transcription
  • ARK-compatible chat providers are now available, including BytePlus, BytePlus Coding Plan, Volcengine Coding Plan, and related OpenAI-compatible ARK endpoints
  • Thinking models can now show reasoning updates while they respond
Changed 3
  • MCP tools are easier to configure and test, with a rebuilt settings page, direct JSON editing, connection testing, server enable/disable controls, and better restart feedback
  • Adjust the character offset in landscape mode for easier positioning of Live2D and 3D characters across desktop, web, and mobile layouts
  • Voice activity settings now expose silence duration and display threshold values correctly
Fixed 4
  • Fixed a critical desktop issue where opening AIRI could briefly flash a blue border around the stage window and leave the inner window unresponsive
  • Model selection is more reliable and keeps the loaded display model and selected character card aligned
  • Plugin tools now fail faster when they cannot respond
  • Missing settings translations were filled in, with broader translation updates across supported languages

From AIRI

New Contributors
v0.10.2
🆕 Updates
New providers
  • You can now use Xiaomi MiMo for speech and transcription.1
  • ARK-compatible chat providers are now available, including BytePlus, BytePlus Coding Plan, Volcengine Coding Plan, and related OpenAI-compatible ARK endpoints.2
Agent capabilities
  • Thinking models can now show reasoning updates while they respond, so long-running answers feel less like a black box.3
  • MCP tools are easier to configure and test, with a rebuilt settings page, direct JSON editing, connection testing, server enable/disable controls, and better restart feedback.4
Interface improvements
  • You can now adjust the character offset in landscape mode, making it easier to position Live2D and 3D characters across desktop, web, and mobile layouts.5
  • Voice activity settings now expose silence duration and display threshold values correctly.6
🐛 Bug fixes
  • Fixed a critical desktop issue where opening AIRI could briefly flash a blue border around the stage window and leave the inner window unresponsive.7
  • Model selection is more reliable. Previously, the loaded display model and the selected character card could fall out of sync; AIRI now keeps them aligned.8
  • Plugin tools now fail faster when they cannot respond, so broken or unavailable tools block the experience for less time.9
  • Missing settings translations were filled in, with broader translation updates across supported languages.101112
✍️ Documentation
  • The setup and usage manual for AIRI Tamagotchi has been refreshed.13

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.10.1...v0.10.2

Footnotes
  1. Commit 5205f369b by @kmiit.

  2. Commit 48d52a2db by @Maaannnn.

  3. Commit e50e140a7 by @felixtremblay.

  4. Commit 81dbd71c3 by @BeanDz.

  5. Commit 85627d34a by @drHuangMHT.

  6. Commit 33cd61111 by @Sambhram1.

  7. Commit e86ee2f13 by @nekomeowww and @Lovehsigure_520.

  8. Commit ef60c157d by @nekomeowww.

  9. Commit 35c3471ec by @nekomeowww.

  10. Commit c6e2f727d by @hahaQWQ.

  11. Commit abc434152 by @github-actions.

  12. Commit 39b119641 by @luoling8192.

  13. Commit 1e80470f0 by @icefair.

View originalPermalink
How v0.10.2 went

v0.10.1

Added 22
  • Sign in with email, verify email address, reset password, and manage profile from the account page
  • Link social accounts to AIRI account for managing different sign-in methods in one place
  • Delete account from settings
  • AIRI Cloud now supports server-side TTS with per-character Flux billing
  • Amazon Bedrock is now available as a chat provider
  • MiniMax Speech is now available as a TTS provider
Changed 4
  • Improved Flux billing reliability for TTS usage, especially when usage needs to be recorded before payment is fully settled
  • Improved token refresh handling to make sign-in state more reliable when multiple requests refresh auth simultaneously
  • Adjusted chat input behavior for smoother typing and message sending
  • Improved WebGPU-based local inference for speech, TTS, and background removal with better model loading, cancellation, and GPU recovery
Fixed 7
  • Chat input flickering and instability issues
  • Several chat layout issues with bubbles, buttons, tabs, and input scrollbars
  • Provider lists in settings that did not scroll correctly
  • Kokoro default model now saves correctly on first visit
  • CJK and keep-text speech processing cases that could strip text incorrectly
  • MacOS cursor, display geometry, iframe coordinate, and cleanup issues in desktop flow
  • Plugin tools not being enabled correctly

From AIRI

v0.10.1 Highlights
Account & AIRI Cloud
  • You can now sign in with email, verify your email address, reset your password, and manage your profile from the account page.
  • You can now link social accounts to your AIRI account, making it easier to manage different sign-in methods in one place.
  • You can now delete your account from settings when you no longer want to keep your data.
  • AIRI Cloud now supports server-side TTS with per-character Flux billing.
  • We improved Flux billing reliability for TTS usage, especially when usage needs to be recorded before payment is fully settled.
  • Some users could run into unstable sign-in sessions when multiple requests refreshed auth at the same time. We improved token refresh handling so sign-in state should be more reliable.
New Providers
  • You can now use Amazon Bedrock as a chat provider.
  • You can now use MiniMax Speech as a TTS provider.
  • Artistry now supports image providers such as ComfyUI, Replicate, and Nano Banana.
Chat & Settings
  • The chat input could flicker or feel unstable in some cases. We adjusted the input behavior so typing and sending messages should feel smoother.
  • When a message fails, you can now retry it directly instead of recreating it manually.
  • Some chat bubbles, buttons, tabs, and input scrollbars did not fit cleanly. We fixed several layout issues to make the chat UI feel more polished.
  • Some provider lists in settings were hard to reach because they did not scroll correctly. We fixed this so all options should be accessible.
  • Kokoro’s default model is now saved correctly on first visit.
  • You can now clear OpenAI-compatible speech model fields when you want to reset them.
Artistry
  • You can now use Artistry to generate or work with images through supported image providers.
  • AIRI Card creation and scene settings now have better image-related controls and previews.
  • Generated image results in chat are easier to inspect with the new journal preview flow.
Models & Local Inference
  • You can now get clearer feedback when importing Live2D models. AIRI checks the model structure and shows a report when something looks wrong.
  • We improved WebGPU-based local inference for speech, TTS, and background removal. This should make local model loading, cancellation, and recovery from GPU issues more reliable.
  • Some CJK and keep-text speech processing cases could strip text incorrectly. We improved the audio text processing pipeline to handle those cases better.
Computer Use
  • AIRI now has an experimental Computer Use foundation for observing the desktop and interacting with browser content.
  • Browser actions are now routed through DOM-aware tools where possible, which should make web interactions more accurate than coordinate-only clicking.
  • We added safer session tracking, read-only DOM inspection tools, transcript-based state tracking, and improved desktop overlay behavior.
  • Some macOS cursor, display geometry, iframe coordinate, and cleanup issues were fixed to make the experimental desktop flow more stable.
Mobile & Stage
  • You can now use a transparent background in Stage Pocket.
  • We added an early Godot Stage preview. This is experimental, but it gives us a path toward richer stage runtimes.
For Plugin Developers
  • You can now build plugins with the new Kits API, Bindings API, Tools API, and Gamelet API.
  • Plugin manifests now use plugin.airi.json.
  • Plugins can now provide custom widget UI.
  • We added a local loopback server so plugin assets can be served more safely inside Stage Tamagotchi.
  • We added an initial chess gamelet example to show how the new Gamelet API works.
  • We improved plugin host structure, debugging, auto reload, asset mounting, and iframe messaging through @moeru/eventa.
  • Some plugin tools were not being enabled correctly. We fixed that so plugin-provided tools can be used as expected.
Developer Tools
  • We added IO Tracer to help inspect how messages, hearing, and other runtime flows move through AIRI.
  • IO Tracer now has better layout behavior, clearer identifiers, and localized route metadata.
  • We added the foundation for global shortcuts.
Upgrade Notes
  • If you self-host AIRI, set BETTER_AUTH_SECRET before upgrading.
  • If you maintain a plugin, migrate your manifest to plugin.airi.json.
  • Artistry, Computer Use, Desktop Overlay, and Godot Stage are still evolving. Some details may change in future releases.
View originalPermalink
How v0.10.1 went

v0.10.0

Pre-release
Added 19
  • Email login and profile authentication
  • OIDC JWT bearer plugin with linked social accounts
  • Account deletion feature
  • Read-only DOM tools parity to extension bridge for computer-use-mcp
  • Transcript truth source and safe projection for computer-use-mcp
  • Extract pure runtime logic from stage-ui into core-agent
Changed 2
  • Require BETTER_AUTH_SECRET and pass it explicitly to betterAuth
  • Add gateway and model configuration to environment variables

From AIRI

New Contributors
🚀 Features
🐞 Bug Fixes
View changes on GitHub
View originalPermalink
How v0.10.0 went

v0.9.0

Added 6
  • Basic MCP implementation for stage modules
  • Popup controls island and better tray with window options for stage-tamagotchi
  • Missing icon for gaming-factorio module in stage-ui
  • Ollama thinking mode config and provider-side think mapping
  • Anti-aliasing support for 3D models
  • Manual model input fallback when model list fails to load in stage-pages
Changed 3
  • Improve controls island tooltip placement in stage-tamagotchi
  • Migrate satori-bot persistence to adapter
  • Migrate minecraft reflex runtime to alien-signals, harden zod validation, and unify skill utilities
Fixed 8
  • Typo in instructions for Chat window
  • Improve VRM expression rendering and add missing emotion mappings
  • Guard against undefined voices in ElevenLabs listVoices in stage-ui
  • Raise ComboboxContent z-index above Dialog overlay in ui
  • Auto-mark provider as added after successful validation on settings page in stage-ui
  • Guard structured message fallback in telegram-bot
  • Surface provider transcription errors in stage-pages and stage-ui
  • Add model list for index-tts-vllm in stage-ui

From AIRI

New Contributors
What's Changed

Full Changelog: https://github.com/moeru-ai/airi/compare/v0.8.5-beta.4...v0.9.0

View originalPermalink
How v0.9.0 went

v0.9.0-rc.1

Pre-release
Added 3
  • New BugReporterDialog component in stage-ui
  • Error container in ui
  • FieldInput, InputFileCard components and story updates in ui and stage-ui
Changed 1
  • Compress and downscale AVIF screenshots in vishot
Fixed 3
  • Position of status icons in stage-pocket
  • stage-tamagotchi no longer unable to build electron due to nested pack
  • Improve LLM KV-cache compatibility for XML context injection in stage-ui

From AIRI

🚀 Features
  • stage-ui: New BugReporterDialog component - by @nekomeowww (b1a4e)
  • ui: Error container - by @nekomeowww (ec4d1)
  • ui,stage-ui: FieldInput, InputFileCard, and story update - by @nekomeowww (14f82)
🐞 Bug Fixes
🏎 Performance
View changes on GitHub
View originalPermalink
How v0.9.0-rc.1 went

v0.9.0-beta.7

Pre-release
Added 3
  • Scan QR code to connect to tamagotchi in stage-pocket
  • Add weather API and tool for widget calling in stage-tamagotchi
  • Let AIRI see tool failures in LLM context with captureToolErrors and xsai patches
Changed 2
  • Adapt QR code to dark mode and move connection component
  • Improve fallback and add matrix smoking test in stage-tamagotchi
Fixed 5
  • Fix outdated import in stage-pocket
  • Fix incorrect auto updater in stage-tamagotchi
  • Fix AutoUpdater with channels in stage-tamagotchi
  • Disable coupon in stage-ui
  • Update RoundRange and correct onboarding props in stage-ui and ui

From AIRI

🚀 Features
🐞 Bug Fixes
  • stage-*:
    • Qr should have dark mode adapted, connection component moved - by @nekomeowww (16017)
  • stage-pocket:
    • Outdated import - by @nekomeowww (92b42)
  • stage-tamagotchi:
    • Incorrect auto updater - by @nekomeowww (d989e)
    • AutoUpdater fix with channels - by @nekomeowww (c2dba)
  • stage-ui:
    • Disable coupon - by @luoling8192 (65419)
  • stage-ui,ui:
    • RoundRange updated, onboarding props corrected - by @nekomeowww (9be9b)
🏎 Performance
  • stage-tamagotchi: Better fallback, matrix smoking test - by @nekomeowww (b95d8)
View changes on GitHub
View originalPermalink
How v0.9.0-beta.7 went

v0.9.0-beta.6

Pre-release
Added 20
  • Add full rollout chat completion check
  • Add browser artifact transformer pipeline
  • Add OIDC authentication
  • Add JWT sign functionality
  • Add tool for interacting with electron apps
  • Add setup and use manual to documentation
Changed 1
  • Move language select to top-right corner
Fixed 5
  • Decouple VRM loading reason and ThreeScene phase transition reason
  • Fix incorrect build for all redistributed fonts
  • Remove duplicate keys in i18n
  • Fix WebSocket cannot reconnect correctly and improve logging
  • Fix missing install of embla-carousel-vue

From AIRI

New Contributors
🚀 Features
🐞 Bug Fixes
  • Decouple vrm loading reason and ThreeScene phase transition reason - by @Lilia-Chen (631a6)
  • ci:
    • Fix missing build-tools - by @LemonNekoGH (b307d)
    • Wrong apk path - by @LemonNekoGH (c1d3c)
  • font-*:
    • Incorrect build for all fonts we re-dist - by @nekomeowww (da405)
  • i18n:
    • Remove duplicate keys - by @LemonNekoGH (4b203)
  • scenarios-stage-tamagotchi-browser:
    • Skip build - by @nekomeowww (d1427)
  • server-*,stage-ui:
  • server-runtime,docs,vishot-*:
    • Tsconfig issue, node issue, h3 export - by @nekomeowww (99e44)
    • Tsconfig issue, node issue, h3 export - by @nekomeowww (73429)
  • stage-*:
  • stage-pages:
  • stage-pocket:
    • Should not reset to default value when entering connections page - by @LemonNekoGH (a7139)
  • stage-pocket-android:
    • Bridge the websocket connection to android native - by @LemonNekoGH (e35b3)
  • stage-tamagotchi:
  • stage-tamagotchi,server-runtime:
    • Crossws error in EADDRINUSE not handled - by @nekomeowww (cb724)
  • stage-ui:
  • stage-ui-three:
  • tests:
    • Server build - by @luoling8192 (457ad)
  • vishot-*,deps:
    • Update lock, incorrect button impl - by @nekomeowww (4880f)
  • vishot-,scenarios-:
    • Incorrect button click impl, outdated import - by @nekomeowww (32c88)
  • vishot-runtime:
    • Missing scale and coordinate relocate - by @nekomeowww (fd50d)
  • workspace:
    • Incorrect font configured - by @nekomeowww (2ba82)
View changes on GitHub
View originalPermalink
How v0.9.0-beta.6 went
View all

Discussion

If you publish AIRI, you can claim this product by proving you administer its repository.