Everything in this release is for Enterprise accounts that connect Reagent to their organization’s Custom Endpoint.
Reasoning Models on Custom Endpoints
OpenAI reasoning models (such as the GPT-5 family and o-series) can now use the tools available on Enterprise, including Wwise questions, when your endpoint offers OpenAI’s Responses API. With API set to Auto, Reagent picks the API for each model after testing your endpoint.More Endpoint Settings
Settings > Endpoint and the Custom Endpoint window now include API, Reasoning effort and, under Advanced, Max output tokens and Context window. In Settings > Endpoint, a Detected row shows which API Reagent uses and why.Improvements
- Apply now tests that your endpoint can handle Reagent’s tool requests before saving, and shows the endpoint’s own error message if it can’t.
- If your endpoint rejects a request, Reagent now retries once in a format the endpoint accepts and remembers what worked.
- The context usage indicator now uses the context window you set for your endpoint (or a default for your model’s family) instead of Reagent’s default.
Bug Fixes
- Fixed Claude responses getting cut short when your endpoint applies a low default length limit.
- Fixed Gemini models failing partway through multi-step tasks when running through your endpoint.
- Fixed Wwise questions failing on endpoints that strictly check Reagent’s tools.
- Errors from your endpoint now show the endpoint’s actual message instead of unrelated advice about Reagent credits.
Claude Sonnet 5
Reagent now runs on Claude Sonnet 5, a newer and more capable model. The Wwise subagent and the Analyst moved to a newer, faster model as well.A New Look
Reagent has a cleaner black-and-white design that matches the website, with redesigned sign-in, settings and onboarding screens, and a new startup animation. Keyboard focus is clearly highlighted throughout, the app follows your system’s reduce-motion setting, and on macOS you can zoom the whole interface from the keyboard.Redesigned Sidebar
Your chats are now grouped by date and searchable, and you can star the ones you keep coming back to so they stay pinned at the top.Collapsible Activity Steps
When Reagent works through several steps, they fold into a single expandable summary instead of burying the answer under a wall of steps. Open it to see what each step did. Steps that return tracks list them by name with their item counts.Sharper Analysis
Loudness measurement is more accurate on bass-heavy material and runs several times faster. Transient detection catches quieter hits in takes that swing between loud and soft passages, and classifying sound content across many tracks now returns a summary — noticeably quicker and far cheaper in credits.Browser Sign-In
Signing in now happens in your web browser, which brings you straight back into the app. It also lets Enterprise accounts sign in with their company’s single sign-on. See Single Sign-On.Custom Endpoint
Enterprise accounts can now connect Reagent to their organization’s own model endpoint. Reagent sends its AI requests there instead of to Reagent’s servers, and doesn’t offer the features that send your content to Reagent’s servers: dialogue transcription, transcript analysis and translation, reference-file uploads, REAPER API search, the Wwise subagent’s online lookups and in-app feedback. Some services are still contacted, such as Reagent’s account service for sign-in and your profile, WorkOS for company SSO, reaperagent.com for release notes, GitHub for updates and PostHog for the analytics library’s settings (no usage events are sent). See where your data goes.Improvements
- Reagent uses fewer credits and responds faster on most tasks, because it now loads only the tools each task needs.
- REAPER edits and analysis now run directly in your conversation instead of being handed off, so tasks finish faster and more reliably.
- When a step fails, Reagent shows a short explanation and, where it has one, a suggested next step instead of a blank line or a wall of error text.
- Reagent now explains when a file type can’t be attached, warns you before you send attachments that are too large, and lets you remove the oversized file without starting over.
- Long pasted text now collapses into a tidy block instead of flooding the message box.
- When REAPER or Wwise isn’t connected, Reagent tells you plainly what it can do and how to connect.
- Wwise scripts now run in an isolated sandbox for added safety.
Bug Fixes
- If part of a longer dialogue analysis fails, Reagent now retries just that part and merges it back correctly.
- Asking for only the complete or incomplete takes in the transcript now filters correctly instead of coming back empty.
- Working with comped takes on fixed lanes now returns the right items even when nothing is selected in REAPER.
- Several REAPER agent actions sent at once no longer interfere with each other.
- Chats that can’t be named automatically now take their title from your first message, and chats stuck on a generic name get a proper title when you reopen them.
- Pasted text now keeps its line breaks.
Already Live on the Web
These reached individual plans (Lite, Pro and Pro+) without updating the app.- You can permanently delete your account and billing yourself from your account page, which also cancels any active subscription.
- Add-on credit top-ups are available on every paid plan, not just the top tier.
Context-Aware Live Wwise Mixing
Reagent now identifies the exact sound variant you just heard playing in your game—regardless of what object you currently have selected in Wwise. For example, if you trigger a dagger swing in-game and say “make that swing 3 dB louder,” Reagent targets the specific dagger container, even if you accidentally clicked on a sword container afterward. Learn moreImprovements
- Improved credit usage for dialogue pipeline.
Performance
Reagent now responds faster thanks to a new agent runtime.Translation PDFs in Non-Latin Scripts
Translation PDFs from the Dialogue Manager now include fonts for Thai, Japanese, Chinese, Korean, Arabic, Hebrew, Hindi and other non-Latin scripts, and align right-to-left languages to the right. Previously these languages came out as Latin-1 lookalikes. Learn moreTool Permission Modes
Choose how often Reagent asks before making changes: Default permissions asks first, Auto-review runs low-risk file changes automatically and asks before riskier ones, and Full access runs changes without approval prompts. Learn moreCancel an Agent Question
When Reagent asks you to choose between options mid-task, you can now dismiss the question with Cancel or Esc and the agent stops cleanly instead of guessing past you. Learn moreImprovements
- Polished approval dialog with clearer presentation of what Reagent is about to do.
- The Windows installer is smaller, so it downloads and installs faster.
Bug Fixes
- Reference scripts (PDF, DOCX, TXT) attached to a chat now properly inform the dialogue pipeline, fixing inaccurate edits when working from a reference.
- Chat content no longer overflows on narrow window widths.
- Resume after approval works reliably across the whole turn.
- Chat session state persists across reloads.
- Cancelled questions no longer keep the timer running.
- Various Windows path handling fixes.
Everything in this release is for the dialogue editor, which isn’t available on Enterprise accounts.
Improvements
- Podcast section mapping better handles off-script content and preambles, preserving the first word of scripted content.
Bug Fixes
- Multi-mic podcast sessions (each speaker on their own mic) now align correctly. Previously one speaker’s track could collapse into a single incomplete take.
- Dialogue cuts now have 20ms edge fades to remove click and step artifacts at midpoint cuts.
- Track and item colors are correctly applied after dialogue editing.
- Various dialogue pipeline reliability fixes.
Slash Commands
Save your most-used prompts as reusable slash commands and trigger them with/ in chat. Build a Wwise audit, a UCS rename routine, or a render pipeline once and run it with a keystroke. Learn moreFile Operations
Reagent can now move, copy, rename, and create folders for you. Every change requires your approval, and a single click can auto-approve the rest of a batch — letting the agent finish the job without prompting on each file. (Since the May 9, 2026 update, permission modes decide when Reagent asks, and the batch option is gone.) Learn moreRender Media
Reagent can now render selected items, tracks, regions, time selection, or the entire project to disk. Built-in presets cover 16-bit, 24-bit, and 32-bit float WAV plus MKV with FFV1 video, with optional sample-rate, channel, and dither overrides. Pairs with the Wwise integration — ask Reagent to render items and drop them straight into your Wwise Originals folder. Learn moreWwise Integration
You can now use Reagent with Wwise. Ask it to help you debug issues, brainstorm different Buss or Container structures, check implementation, duplicate objects with same settings and properties etc… This integration works both on Windows and macOS and supports any Wwise version starting from 2021.1. Learn moreImprovements
- Subagents now work more efficiently with less credit usage.
Bug Fixes
- Fixed tool names in chat.
Translation for Non-English Audio
Reagent now detects the language during transcription and offers an English translation for non-English recordings. Translations appear in a dedicated Dialogue Manager viewer with seek-to-line playback and PDF export. Learn moreMulti-Item Podcast Analysis
Podcast mode can now analyze multiple items as a unified entity, producing more accurate alignment and edit suggestions for recordings split across several items.Improvements
- Voice-takes mode now automatically reconstructs a reference script when none is provided, significantly improving take classification accuracy.
- Reagent can now directly edit media items without delegating to a background task.
- Timer pauses when waiting for your input instead of continuing to count.
- When Reagent asks questions and app is not focused, desktop notification will be shown.
- Improved dialogue pipeline for longer recordings with server-side chunking.
- Reagent is now lane aware, therefore if you have a track with multiple lanes, it will automatically use the comp lane by default.
- More improvements to credit usage for dialogue pipeline.
Bug Fixes
- Fixed rate limit messages incorrectly showing as quota exhaustion.
Podcast / Narration Style Suggested Edits
The podcast pipeline now shows suggested cuts as red (incomplete) items in REAPER, letting you preview what will be removed before committing. Learn moreImprovements
- Overall dialogue pipeline accuracy improvements, especially on longer media items.
- Dialogue analysis now asks you to choose between podcast and voice-takes modes for more reliable results.
- Improved voice-takes accuracy with clean script preprocessing before analysis.
- Improved dialogue pipeline when no script is provided.
- Dialogue pipeline works on video media items, so you can have Reagent edit videos for you.
Bug Fixes
- Fixed overlapping takes in dialogue pipeline after processing.
- Fixed an issue where text-based reference files could fail to load.
REAPER Subagent
Reagent can now delegate complex multi-step tasks to a specialized REAPER subagent. This was implemented to reduce credit usage and improved performance/accuracy when working on complex tasks. You’ll see a live activity view in the chat as it works through each step — all changes remain undoable in REAPER. The REAPER subagent was removed in the September 16, 2026 update; REAPER work now runs directly in your conversation. Learn moreCopy Chat
Copy your entire conversation to the clipboard with one click — useful for sharing workflows or sending feedback. Learn morePDF & Document Uploads
Attach PDF documents, Word files (.docx), and Excel spreadsheets (.xlsx) directly in your chat alongside existing file types. On Enterprise accounts, Word and Excel files can’t be attached. Learn moreImprovements
- Your chat input is now kept when you switch sessions, so a message in progress is still there when you come back (until you quit Reagent).
- Improved responses with better understanding of REAPER tools and when to use them.
- Extended timeouts for longer-running operations to prevent interruptions during complex tasks.
- Reduced memory usage through more efficient project snapshot handling.
- Improved credit usage, tasks taking fewer credits to execute.
Faster execution
Reagent now executes faster and more efficiently, thanks to the new snapshot tool. This helps Reagent use fewer credits as well.Dialogue Pipeline Upgrade
Dialogue editing is now more flexible, allowing it to follow custom instructions for non-standard workflows like speaker detection or topic segmentation. E.g. “Take the selected item. Anytime you hear a person’s name, slice to create an item and color it red.” or “Edit VO and add marker where the speech differs from the script.” etc… Learn moreFile Attachments in Chat
Attach reference files directly in the chat input — scripts, spreadsheets, PDFs, and more. Reagent uses the file during processing. On Enterprise accounts, Word and Excel files can’t be attached and reference files aren’t used for dialogue editing. Learn moreBetter REAPER context detection
Reagent now reads your current REAPER state — selected items, tracks, cursor position, and time selection — every time you send a message. This helps it understand what you’re referring to. Learn moreAudio Classification for More Formats
Sound classification now works with common audio formats such as WAV, MP3, FLAC, OGG and AIFF, not just WAV. (On a Mac, classification needs Apple silicon and macOS 14 or later. See System Requirements.) You can even ask it “find track containing VO” or anything along those lines.Create Variation Tool
Creating variations is now much faster, and you can chain prompts together:Improvements
- Chat titles are generated instantly from your first message.
- Added Copy button on user messages.
- Reagent now writes less filler and acts more directly, which uses fewer credits.
- Improved tool selection so the agent picks the right tool more consistently.
Bug Fixes
- Fixed incorrect chat messages rendering.
- Fixed the splash screen not being draggable.
Bug Fixes
- Fixed an issue where audio processing features could fail on Intel-based Macs.
Cloud-Powered Dialogue Editing
Transcription and analysis now run in the cloud for faster, more accurate results — no more downloading local models or waiting for on-device processing. Learn moreIncomplete Take Preservation
False starts, mid-word cutoffs, and interrupted takes are now kept in the edit and colored red in REAPER, giving editors full visibility into all delivery attempts.Improvements
- Dialogue editing results appear in REAPER as each item finishes processing instead of waiting for the entire batch to complete.
- Progress indicators now show detailed stages during dialogue editing instead of a single generic message.
- Context usage indicator now persists across sessions.
Bug Fixes
- Fixed media items not being automatically renamed to their spoken text after dialogue editing.
- Fixed an issue with app redirect after authentication.
LUFS Loudness Targeting
The Take Volume Leveler now supports LUFS targeting. Set a specific loudness target like -24 LUFS for broadcast and Reagent measures and adjusts automatically. Learn moreWhat’s New Dialog
After updating, you’ll see a dialog showing exactly what changed in the new version — no more guessing what’s different.In-App Feedback
Send bug reports, feature requests, or general feedback directly from the app without leaving your session.Context Usage Indicator
A ring next to the send button shows how much of the conversation window you’ve used. It turns amber at 80% and red at 95% so you know when to start a new session.Improvements
- Download progress ring in the title bar shows real-time update download progress.
- Actionable toast notifications with buttons like “Install Now” let you respond to events without disrupting your workflow.
- Redesigned chat input layout with better visual hierarchy and a toolbar row below the text field.
- Take Volume Leveler now shows descriptive status messages during processing.
- Faster performance with large projects — media items now load in pages instead of all at once.
- Added a changelog link in Settings for quick access to release notes.
- Longer running sessions will no longer throw authentication errors.
Improvements
- Transcription processing is now offloaded to a dedicated worker for better performance and stability.
- Chat messages now show a turn timer so you can see how long each response takes.
- Improved progress tracking during batch tool operations.
Auto Dialogue Editor Upgrades
Voice take splitting now handles full recording sessions — 30 minutes, an hour, even 2+ hours. Longer recordings are automatically broken into segments and processed in parallel. You can also provide a script reference file (CSV, TSV, TXT, or XLSX) to significantly improve take classification, and edited items are now renamed to match what was actually said. Learn moreReaScript Authoring
Reagent can now create, save, and register custom Lua scripts as REAPER actions — all from the chat. It can run a script right away (a Run script block) or draft one for you to review first (a Create script block). Scripts can be assigned keyboard shortcuts on first save, and Reagent is aware of scripts in your Scripts directory so you can ask it to update or extend them. (Since the May 9, 2026 update, Reagent asks before running a script unless you use Full access. See Permissions.) Learn moreImprovements
- Bridge update notices: when an app update changes the Bridge, Reagent updates the files and asks you to restart the Bridge script in REAPER.
- Script save names now auto-sync with the description for easier identification.
Voice Take Splitting
Reagent can now automatically split dialogue and speech recordings into individual takes. It uses local speech recognition (Whisper) running on your machine. (Since the March 1, 2026 update, transcription runs in the cloud.) It detects take boundaries using word-level timestamps, trims silence, and edits directly in REAPER. Learn morePro+ Plan
For heavy users, there’s now a Pro+ plan with more monthly credits than Pro. See current plans at reaperagent.com.Improvements
- Credit balance dashboard — you can now see your credit balance directly in your dashboard.
Bug Fixes
- Fixed an issue affecting bridge connections on Windows.
Improvements
- New trial credits system — new users now get trial credits on signup for simpler onboarding.
Take Volume Leveler
The Take Volume Leveler automatically analyzes your dialogue or vocal tracks and generates volume envelope automation to level out the dynamics. Select your media items and Reagent handles the rest — analyzing amplitude, calculating gain adjustments, and applying smooth, natural-sounding volume automation directly to the take envelope. All changes are non-destructive envelope automation, so your audio stays untouched. Learn moreImprovements
- Improved FX parameter handling for more reliable plugin control.
- Enhanced media item data retrieval with additional track information.
- Real-time subscription tier updates now sync instantly in the app.
Improvements
- Improved markdown rendering in chat messages.
- Reduced unnecessary logging for a cleaner experience.
- Better handling of ambiguous requests with guided context gathering.
Subscriptions & Credits
Reagent now has subscription plans with credit-based usage and a real-time credit balance display. Your credits update instantly as you use Reagent, and you’ll see a clear message when your quota is exhausted.UCS Audio Classification
Classify your sounds using the Universal Category System with a trained model built into Reagent. Identify sound types and categories directly from your REAPER project.File Attachments
Drag and drop files or folders into the chat, or use the file picker to attach reference material for your requests.Improvements
- Delete chat sessions with right-click context menu. (Since the September 16, 2026 update, delete chats from the ⋯ menu next to each chat in the sidebar.)
- Clear all conversations from the sidebar.
- Window position and size are now remembered between sessions.
- Improved onboarding flow with dedicated steps for SWS Extensions and macOS permissions. (Onboarding has since changed. SWS and Bridge setup now live in Device Setup, which opens from the Setup button on the “REAPER bridge not installed” banner. See Quickstart.)
Bug Fixes
- Fixed traffic light button positioning on macOS.
- Fixed skills not being recognized in certain configurations.