AI Dev Kit v4

4.8.0 (2025-11-30)

Fixed:

• Model and Voice assets path normalization issues resolved

---

4.7.29 (2025-11-29)

Fixed:

• Unity 2022 backward compatibility restored

---

4.7.28 (2025-11-28)

Added:

• DeepSeek API support (Pro package)

Fixed:

• Runtime path normalization issues

• Editor path normalization issues

• Transcriber component issues

---

4.7.24-4.7.27 (2025-11-26 ~ 2025-11-28)

Added:

• GroqCloud file support in Data Manager

• GetCredits() request - returns prepaid balance (used/total)

• FileMetadata to file classes

Fixed:

• Multiple filtered popup issues (#2, #3, #4)

• Path issues resolved

Removed:

• Removed unused 'RESEARCH_LAB' script define

---

4.7.23 (2025-11-23 ~ 2025-11-24)

Added:

• OpenRouter Files API (Data Manager)

• Voice GetVoice() extension method

• GENSpeech SetVoice() extension method

• GENSpeech SetModel() extension method (non-generic)

• Model GetModel() extension method

• Embedding Request SetModel() extension method

• Get Voice API (Google Gemini, ElevenLabs)

• File upload / list / delete requests (OpenRouter)

Fixed:

• Voice Metadata Generator / Model Metadata Generator Window (error popup fixed)

• Streaming handler resolved

• Streaming Audio Player fixed

• Auto-resolve PlaybackFormat from request format

• Filtered Popup search function fixed

• Generated Voice OnPronounce() callback fixed

Changed:

• Dev Chat tool type changed from Coding Action -> Text Chat

---

4.7.22 (2025-11-21)

Added:

• Perplexity API support (+OpenRouter)

• xAI API support (+OpenRouter)

• Cohere API support (+OpenRouter)

• Mistral API support (+OpenRouter)

• OpenRouter API support (OpenRouter)

• OpenRouter Files API support

• New model metadata system

• Voice metadata system

• Research Lab tier (Pro users exclusive, preview feature)

Implemented:

• Chat request with tool calls (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Chat request without tool calls (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Tool calling (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Tool conversion (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Streaming response (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Streaming tool calls (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Input audio (Cohere, Mistral, OpenRouter, Perplexity, xAI)

• Multiple system messages (Cohere, Mistral, OpenRouter, Perplexity, xAI)

Changed:

• Provider naming: ElevenLabs Audio API -> ElevenLabs

• Provider naming: OpenRouter (Text Only) -> OpenRouter

Fixed:

• OpenAI Realtime Metadata (event id, item id, etc.)

• Tool Events now properly pushed to Debug Window

---

4.7.20 (2025-11-19)

Added:

• Duplicate Model / Voice feature in Model Library & Voice Library tools

• Model / Voice preset management system

• Voice preset management UI

Changed:

• Data Manager: Files are now represented as Unity assets instead of pure C# classes

• FileManager class changed to FileCatalogue

• File assets changed to class File where T can be AudioClip, Texture2D, or VideoClip

---

4.7.18-4.7.19 (2025-11-16 ~ 2025-11-18)

Added:

• Multiple Tool result events

• Input audio format handling

• Output audio format handling

• Anthropic Message Batches API

• OpenRouter File API (untested)

• Workspace window for navigating editor tools

Fixed:

• File API issues

• GenerateSpeech issues

---

4.7.16 (2025-11-11 ~ 2025-11-13)

Added:

• File manager tool (File API management tool)

• DLL missing warning window at Editor start

• Anthropic Prompt Caching API

• OpenAI GPT-4o Audio Preview API

• Google Gemini Context Caching API

Fixed:

• Fixed input delay for speech transcriber

• Fixed Speech Generator double slash issue

• Improved Google Gemini client error message

---

4.7.10-4.7.11 (2025-11-09)

Added:

• Tool error messages to Debug/Log window

• Tool failure events

• Utility class to get model capabilities at runtime

• OpenAI Responses API Draft (preview, limited testing)

Changed:

• Tool output now logged to Debug/Log window (tool name shown for better readability)

---

4.7.7-4.7.9 (2025-11-07 ~ 2025-11-08)

Added:

• JSON Schema tool description support

• Google Gemini grounding feature with Google Search

• Google Gemini grounding feature Code Execution

• Local Shell Tool support (Windows/Mac/Linux command execution)

• Google Gemini audio generation API

Changed:

• Simplified GEN Tasks

Fixed:

• OpenAI Realtime API audio output issue

• Multiple Tool-related errors

---

4.7.7 (2025-11-05)

Added:

• Tool Definition with Google Gemini support

• Tool calls from AI Responses with Google Gemini

• Server => Client tool confirmation (supported on OpenAI, Google Gemini)

• Server => Client request data (supported on OpenAI)

• MCP Server support (stable version)

• Google Gemini audio input support

• Google Gemini video input support

• Google MakerSuite files API (untested)

Removed:

• Obsolete Anthropic client APIs

Fixed:

• Various minor issues

---

4.7.6 (2025-11-03)

Fixed:

• Speech Generator audio caching

• Fixed Voice Changer issue

---

4.7.5 (2025-10-29)

Added:

• OpenAI Realtime API – text/audio streaming support

• OpenAI Realtime API – client => server function calls

• OpenAI Realtime API – server => client function requests

• OpenAI Realtime API – audio response streaming (input/output formats selection)

• OpenAI Realtime API – assistant voice customization

• OpenAI Realtime API – audio input/output event hooks

• OpenAI Realtime API – conversation item serialization

• Assistants API – thread run support

• Assistants API – streaming support

• Assistants API – tool output submission

Fixed:

• Model Library issues

• Multiple demo scenes issues

---

4.7.4 (2025-10-27)

Fixed:

• Model / Voice Library changes not reflecting in prefabs/scenes

---

4.7.2 (2025-10-26)

Fixed:

• ScriptableObject Toolkit initialization issues

---

4.7.1 (2025-10-24)

Added:

• Auto-conversion to .wav format if format not supported

• Editor tooltips for model library fields

Fixed:

• GENTask errors

• Voice metadata generator

• Model metadata generator

---

4.7.0 (2025-10-23)

Added:

• Lyria Music Player component

• Lyria Music Track (ScriptableObject)

• Google tools: Google Search, Code Executor, URL Context

• OpenAI Streaming Image API

• TextEmbeddingTask (.GENTextEmbedding())

v.4.6.5 (Sep 21, 2025) – Preview

• Implemented Responses API

• Implemented MCP Tools

• Implemented Local Shell Tools

v.4.6.2 (Sep 2025)

• Internal unitypackage build (40602) / packaging updates

v.4.5.32 (Sep 03, 2025) – Pre-Release

• Added Voice Changer Window (new Generator Window)

• Upgraded Generator Window UIs

• Fixed Assistant Profile inspector issues

v.4.5.27 (Aug 30, 2025)

• Fixed missing marker issues

• Fixed initial setup issues

• Cleaned up initial setup process

v.4.5.13 (Aug 22, 2025) – Pre-Release

• Fixed assistant chatbot demo scene

• Fixed realtime chatbot demo scene

• Upgraded MonoBehaviour and ScriptableObject custom editors

v.4.5.11 (Aug 22, 2025)

• Fixed broken sample scenes

• Upgraded model asset inspector

v.4.5.09 (Aug 22, 2025) – Hotfix

• Generator Window hotfixes

v.4.5.06 (Aug 22, 2025) – Pre-Release

• Merged Pro assembly into Core (simpler setup)

• Added Audio Clip–based Prompt Generator window

• Fixed streaming chat not pushing response messages

• Added Editor session info formatter to Anthropic chat requests

• Polished & stabilized GenAI Windows

• Verified Unity 2022 compatibility

v.4.2.22 (Aug 10, 2025) – Preview

• Voice Library GUI updates

• Voice Metadata system upgrades

• New Locale management system (replaces SystemLanguage, stays compatible)

v.4.2.21 (Aug 10, 2025) – Preview

• Test implementation of Microsoft Azure TTS/STT (experimental)

v.4.2.20 (Aug 08, 2025) – Preview

• Applied Chatbot/Assistant naming rules

  • Components: Chatbot (e.g., Realtime Assistant → Realtime Chatbot)

  • Profiles: Assistant Profiles

• Removed icons from foldout headers

• Various editor GUI refinements

v.4.2.19 (Aug 07, 2025) – Hotfix

• Cleaned up inspectors

• Fixed model ID issue in usage tracking (ChatSession)

v.4.2.18 (Aug 07, 2025) – Preview

• Introduced Generative Profile System

• Added ScriptableObject assets:

  • Local Chatbot Profile

  • Assistants API Chatbot Profile

  • Realtime Assistant Profile

• Implemented custom inspectors for new profiles

• Renamed: VoiceGender → Gender, Chatbot → LocalChatbot, ChatbotBase → Chatbot, Moderator → ContentModerator

• Enhanced ScriptableObject debugging tools; removed redundant editor code

v.4.2.17 (Aug 06, 2025) – Hotfix

• Added fallback codes for ProjectContext (ScriptableObject)

v.4.2.16 (Aug 05, 2025) – Preview

• Added Artificial Analysis API (internal)

• Refactored model metadata resolvers

v.4.2.15 (Aug 06, 2025) – Hotfix

• Fixed chat profile not applying to Chatbot

v.4.2.14 (Aug 05, 2025) – Hotfix

• Added fallback codes to Chatbot, ChatSession, ChatSessionController

v.4.2.13 (Aug 05, 2025) – Hotfix

• Fixed prompt checks

v.4.2.12 (Aug 04, 2025) – Preview

• Added Realtime Assistant Profile

• Refactored Realtime Assistant

• Refactored Chatbot (Assistants API)

• Removed legacy Assistant lifecycle events

• Added new Assistant status event

• Added Conversation Management support

• Redesigned FileManager editor UI

• Added Assistant Thread TreeView editor window

v.4.2.11 (Aug 01, 2025) – Preview

• Refactored UnityEvents & custom inspector handling

• Improved UnityEvent serialization & GUI rendering

• Added editor icons

• Renamed collections: ReferencedDictionary → SerializableDictionary, Metadata → SerializableMetadata, Database → SerializableDatabase

v.4.2.10 (Jul 29, 2025) – Preview

• Continuing Pro sample fixes

• Removed unused editor icons

• Organized editor icons & GUI labels

v.4.2.9 (Jul 29, 2025) – Preview

• Finished new inspector designs

• Fixed WebGL build issues (not fully tested)

• Added/replaced editor icons

v.4.2.8 (Jul 28, 2025) – Hotfix

• Minor fixes ("dumb mistakes")

v.4.2.7 (Jul 28, 2025) – Preview

• New inspector designs (almost done)

v.4.2.6 (Jul 28, 2025) – Preview

• Working on new inspector designs (not finished)

• Separated Speech Recorder from AI components that use recording

v.4.2.5 (Jul 28, 2025) – Preview

• WIP: new Inspector designs (not finished)

• Speech Recorder separated from AI components using recording feature

v.4.2.4 (Jul 27, 2025) – Preview

• Refactored Chatbot & ChatSession architecture

  • ChatSession now holds mutable data only

  • Introduced ChatbotProfile (ScriptableObject) for immutable data (Create > AIDevKit > Chatbot)

  • Chatbot requires both ChatbotProfile + ChatSession

• Added Custom Inspectors for Voice & Model assets

• Added fallback logic if Default Resources aren't generated

v.4.2.3 (Jul 25, 2025) – Preview

• Fixed new chat issue in EditorChat

• Moved Styles into Gizmos folder

• Improved Generator Window UI

• Improved Hub Window UI

• Tested & debugged GroqCloud TTS

v.4.2.2 (Jul 24, 2025) – Preview

• Added GroqCloud integration (Chat tested; STT/TTS untested)

• Improved model & voice metadata management

• Redesigned GENTask option system → .SetOptions(TOption options)

  • Example: .SetOptions(new OpenAICompletionOptions())

• Added temperature, seed, stopSequences to GENResponse & ChatSession

• Removed ReasoningOptions, WebSearchOptions, AdvancedModelSettings, SpeechOutputOptions → migrated into provider-specific option classes

v.4.2.1 (Jul 23, 2025) – Preview

• ApiSettings are no longer singletons → access via clients

  • Example: OpenAI.DefaultInstance.Settings (instead of OpenAISettings.Instance)

• New IApiUser interface for in-game users

  • Provides Id + GetApiKey(Api api) override

• Improved UI for AI Dev Kit Hub (Explorer)

• Refactored ApiSettings architecture

• Added dedicated custom editor for .asset files

v.4.2.0 (Jul 22, 2025) – Preview

⚠️ Super Warning: Remove previous version completely (many GUID breaks).

• Complete cleanup of project + strict assembly management

• Added GENPixelArt to fluent API

• API-specific enums renamed (e.g., OpenAI.ImageDetail, GoogleTypes.PersonGeneration)

• Added CancelLastRequest(), CancelAllRequests() to AIComponent

• Project Context settings separated from AIDevKitSettings → standalone Editor Window

v.4.1.3 (Jul 18, 2025) – Preview

• New UnityEvent Wrappers & Improvements

  • StreamingUnityEvent<> wraps 3 UnityEvents: OnStartStreaming, OnReceiveData, OnFinishStreaming

• Improved CustomPropertyDrawer (new collapsed texture)

• Renamed Components:

  • AIComponentBase → AIComponent

  • GeneratorComponentBase → AIContentGenerator

  • AudioRecordingComponentBase → SpeechToContentGenerator

v.4.1.2 (Jul 17, 2025) – Preview

• Added Api parameter to GENTasks

• Added stricter Api checks

• Added new UnityEvent system (replace chat event receivers)

• Restored demo scenes (not tested)

• Added storage location selection to Chat Session

• Added automatic Chat Session creation on Chatbot start

v.4.1.1 (Jul 16, 2025) – Preview

• Component reworks completed

• ExTreeView reworks completed

• Added attachment transfer in GENStructTask

• Fixed broken demo scenes

v.4.1.0 (Jul 15, 2025) – Preview

• Complete rework of component architectures & custom editors

• Model/Voice Library abstractions (unstable)

• Removed UniTask async methods with return values from components

• Renamed Components:

  • TextToSpeech → SpeechGenerator

  • SpeechToText → SpeechTranscriber

• Renamed Classes:

  • Content → ChatContent

  • ContentPart → ChatContentPart

  • RequestType → RequestTypes

• Removed IChatbot

v.4.0.9 (Jul 14, 2025) – Preview

• Chatbot inspector rework (in progress)

• Added task queue handler to all components (not tested)

• Added Shader Generator windows

v.4.0.8 (Jul 13, 2025) – Preview

• Prompt history saved as JSON (instead of ScriptableObject)

  • Now trackable on devices and at runtime

  • More stable than ScriptableObject (slightly slower)

• Minor Generator Windows fixes

v.4.0.7 (Jul 13, 2025) – Preview

• Fixed File management system

• Added Demo: Chatbot + Speech-to-Text

v.4.0.6 (Jul 11, 2025) – Preview

• Refactored ChatSession for performance

• Added autoStopRecordingOnSilence option to AudioRecordingAIComponent

• Updated RealtimeAudioRecorder

• Fixed modality flag dropdown in Model Library

• Renamed Classes:

  • ChatCompletionStreamHandler → StreamingChatHandler

  • ChatCompletionChunk → StreamingChatChunk

  • AIServerSentEventHandler → ServerSentEventHandler

  • AIServerSentEvent<T> → ServerSentEvent<T>

  • AIServerSentEventBuffer<T> → ServerSentEventBuffer<T>

v.4.0.5 (Sep 21, 2025) – Preview

• Implemented Responses API

• Implemented MCP Tools

• Implemented Local Shell Tools

v.4.0.4 (Aug 30, 2025) – Preview

• Fixed missing marker issues

• Fixed initial setup issues

• Cleaned up initial setup process

v.4.0.3 (Aug 22, 2025) – Pre-Release

• Fixed assistant chatbot demo scene

• Fixed realtime chatbot demo scene

• Upgraded MonoBehaviour components custom editors

• Upgraded ScriptableObject assets custom editors

v.4.0.2 (Aug 22, 2025) – Pre-Release

• Merged Pro assembly into Core to simplify setup

• Added Audio Clip–based Prompt Generator window

• Fixed streaming chat not pushing response messages

• Added Editor session info formatter to Anthropic chat requests

• Polished & stabilized GenAI Windows

• Verified Unity 2022 compatibility

v.4.0.1 (Aug 10, 2025) – Preview

• Test implementation of Microsoft Azure TTS, STT

• Voice Library GUI updates

• Voice Metadata system upgrades

• New Locale management system (replaces SystemLanguage, backward compatible)

v.4.0.0 (Aug 7, 2025) – Preview

• Introduced Generative Profile System

• Added Local Chatbot Profile, Assistants API Chatbot Profile, Realtime Assistant Profile

• Implemented custom inspectors for new generative/chatbot profiles

• Renamed:

  • VoiceGender → Gender

  • Chatbot → LocalChatbot

  • ChatbotBase → Chatbot

  • Moderator → ContentModerator

• Enhanced ScriptableObject debugging tools

• Removed redundant editor code