Read Link and Hub Preview Tabs

TL;DR: Alter can fetch and read web pages with the new Read Link tool and Hub tabs now open in VS Code-style preview mode, and a other quality-of-life fixes landed from community feedback.

New Features & Enhancements

• Read Link tool: New default LLM tool for fetching a URL. Supports plain text, Markdown, HTML, PNG/JPEG/WebP images, and PDF/DOCX/XLSX documents.
• Hub preview tabs: Hub tabs now open in a VS Code-style preview mode. Preview tabs are replaced when opening another item.
• Generate title: Right-click on chats and meetings from the Hub sidebar context menu and click Generate titles to regenerate from their context.
• New Models: Gemini 3.8 flash, Cerebras Qwen 3.8 27B, Qwen 3.8 Flash, GLM 5.3 and GLM 5.3 Flash

Bug Fixes & Stability

• Task tool: tools is no longer required.
• Coding tools: New coding-tool sessions default their working directory to the current workspace directory.
• Title generation: Fixed an issue where titles and meeting short summaries were not being generated
• Conversation links: Safe local file:// URLs are allowed where appropriate in conversation content.
• Quick Hub: Faster text selection capture when invoking quick actions.
• Quick actions: Originating presentation closes correctly after use.
• Straico: Removed the integration.

Grok and Pi Coding Tools, Plus Scheduled Gateway Hours

TL;DR: Run Grok and Pi as local coding tools from Alter, connect Grok as a custom provider via subscription OAuth, and schedule when your gateways are available with automatic wake.

New Features & Enhancements

• Grok coding tools: Run and manage Grok coding tasks from the Hub with a full local tool integration.
• Pi coding tools: Add Pi as a local coding tool you can launch and monitor from Alter.
• Grok custom provider: Connect Grok as a custom model provider via subscription OAuth in Settings.
• Scheduled gateway hours: Gateways can follow an open-hours schedule, waking automatically when availability starts.
• Transcription replacements: Custom dictation replacements now apply across speech workflows, not just live dictation.

Bug Fixes & Stability

• OpenCode: Standalone session command works correctly.
• ChatGPT / Codex: Detached task continuation and async coding task results in gateways are handled reliably.
• Discord: Improved chat resume context when picking up gateway conversations.
• Gateways: Sleep is restored when availability stops.
• Hub: Empty composer switchers dismiss correctly; single-character action search works.
• Quick actions: Batch deletion is supported.
• Dictation: Custom replacements are more reliable.

Manage Your Coding Agents in Alter

TL;DR: Manage your coding agents in Alter — run and monitor Cursor, Claude Code, Codex, ChatGPT, and OpenCode sessions from the Hub and QuickHub.

Video:

New Features & Enhancements

• Local coding tools: Integrate Cursor, Claude Code, Codex/ChatGPT, and OpenCode as local session tools you can launch and control from Alter.
• Details bar: The right sidebar has a new design and section to display ongoing tasks, artifacts, sources and tool call details.
• Composer context: The composer shows which action is currently in use.
• Export chat frontmatter: Exported chats include frontmatter with the model used.

Bug Fixes & Stability

• Localization: Refreshed translations for French, Spanish, Italian, Portuguese, German, Simplified Chinese, and Traditional Chinese.
• Hub: Attach transcript context before send, preserve sources after sending, focus workspace search on popover open, defer tab context menus until hover popover teardown, and align sidebar section padding with list items on macOS 26+.
• Action Editor: Disable Tools switch state persists across section changes; section double-click no longer opens onboarding.
• Conversation view: Hide turn index for single-turn conversations.
• Streaming: Surface HTTP errors during chat streaming.
• Transcription: Harden model readiness checks.
• Menubar: Unclip the first status-item menu row.
• Share file: URI-decode paths when % characters are present.

Finding What You Need Faster

TL;DR: Search inside an open conversation, jump through longer Hub chats, search across your conversation history, and narrow results by conversation type, workspace, action, or tag. Action Editor filters are updated too.

Video:

New Features & Enhancements

• In-conversation search: Search inside the open conversation with ⌘F, then jump between matches with ⌘G and ⇧⌘G.
• Hub history search: Search across all conversations in your Hub history with ⇧⌘F.
• Turn navigation: Jump through longer Hub conversations with a turn index, so you can move between messages without losing your place.
• Hub filters: Narrow the sidebar by conversation type, workspace, action, or tag.
• Action Editor filters: Updated sidebar filters, including folder filter shortcuts.
• RTL conversations: Conversations can display in a right-to-left layout.
• Reasoning selection: Your model reasoning choice is remembered across conversations.
• Scheduled gateways: Scheduled actions can route to a specific gateway destination.

Bug Fixes & Stability

• QuickHub: Stage Manager attachments stay in place.
• Text Selection Capture: More reliable behavior, especially in Google Docs.
• Conversation view: Tool-call glimmer is consistent; the conversation stays above the composer.
• Transcription: Imported audio persists more reliably.
• Meetings: Context is not sent when there is no transcript file.

Choose How Much Alter should think

TL;DR: Thinking Effort puts you in control of model performance — choose between low, balanced, or max effort to match your specific task, and Alter will remember your preference for every model. Reasoning context is preserved across follow-up turns, ensuring a seamless experience whether you need lightning-fast responses or deep analytical depth.

Video:

New Features & Enhancements

• Thinking effort: Choose each model’s advertised thinking effort from Hub, QuickHub, or the Action Editor. Alter remembers the choice per model and uses it for future chats and actions.
• Dictation: Improve dictation speed for parakeet processor.
• Custom Endpoints: Context Window size, thinking and vision capabilities are detected automatically for LM Studio, OpenRouter, Ollama, Bifrost and Gemini.
• Action Editor: A new searchable model picker groups the current model, favorites, and other models, with controls for thinking effort, context, image support, and favorites.
• Dictation shortcuts: Dictation now accepts modifier-only shortcuts, including Command, left or right Control, and left or right Option.
• Discord sources: Web search sources are formatted more clearly in Discord responses.
• New models: New models: GLM-5.2 and deepseek v4 flash 0731

Bug Fixes & Stability

• Model selection: Recover automatically when a saved model is no longer available, align switcher metadata consistently, and clarify vision support without redundant selection indicators.
• Tools: Code Runner now shows its sandbox tool.

Control Alter from your phone

TL;DR: Gateways are coming to Alter. It lets you connect Discord and Telegram, so you can send messages, record voice notes, answer follow-up questions, retrieve files, and run your existing Alter actions — all while Alter runs locally on your Mac.

Important: Gateways are connections to Alter running locally. Alter must remain open on your Mac. This is not a cloud-hosted Alter instance.

Video:

New Features & Enhancements

• Gateways: Multiple Discord and Telegram instances; Discord conversation threads with interactive buttons for access prompts; Ask Anything on Telegram; voice-note transcription in gateway context; Markdown tables rendered for Discord; availability controls to stay online with the lid closed; per-gateway action availability and access settings; doctor checks with docs links; localized gateway settings and Action Editor help.
• Workspace document cache v2: Faster, more reliable workspace document caching for attachments.

Bug Fixes & Stability

• Hub & QuickHub: Shortcut conflicts with split panels; attachment shortcuts route reliably; workspace attachment context and presentation preserved.
• Dictation & transcription: Long transcription results no longer truncated; Argmax custom-vocabulary race during in-flight transcribe; microphone continuation leak in transcript-ready prompt.
• Fair use: Reset countdown uses provider UTC reinit date when available; compile fix for missing fair-use reset countdown.
• Other: MiniMax M3 Flow parameter fix; sandbox container cleanup.

Hub Split Panel

TL;DR: Split Panel lets you add a second panel next to your current Hub session — work in parallel on two tasks or one complex project without losing your place. Drag text between panes, move tabs side to side, @-reference the other panel, and open Markdown, HTML, or PDF files in a side panel with “Open in the other panel.”

Video:

New Features & Enhancements

• Hub Split Panel: Add a second panel next to your Hub session to work in parallel — with shortcuts and stable new-chat / drop behavior.
• Drag-and-drop context: Select text in one session and drag it into the other panel to ask a question or add it to your task.
• Flexible tabs: Reorder tabs or move a tab from one side panel to the other (including on external displays).
• Cross-panel referencing: Use @ to refer to and add context from the other open panel.
• Integrated file viewer: Open Markdown, HTML, and PDF files in a side panel via “Open in the other panel” — inspect documents next to Alter without a third-party editor.
• Audio workflows: Microphone and transcription run as workflows — queued transcription, workflow status in the mic menu and Edge Bar, and local speech diagnostics to help us debug users.
• Shell tools: Shell commands can include a purpose for clearer tool calls.
• ChatGPT tool calls: Tool calls display as they arrive during streaming.
• Localizations: Hub suggestion cards and new split-pane UI strings localized for non-English users.

Bug Fixes & Stability

• Hub: Blank window presentation, QuickHub hosting constraint loop, model sync feedback loop, overlay inspector with split layout, and input shortcuts on macOS 27.
• Conversation: Preserve generation errors; send queued messages after tools finish.
• Audio / transcription: Invalid microphone devices, meeting input picker flicker, hardened voice workflow processing, and rejection of malformed transcript timestamps; skip corrupt transcript rows during conversion.
• Custom endpoints: Better compatibility with exotic OpenAI-compatible endpoints.
• Editor: Prevent Sonoma sidebar truncation.
• Menubar: Clicks restored on macOS 27; workflow status shown in the menu.

Memory

TL;DR: Memory is now first-class in Alter — scoped memory captures what matters automatically, a Catch-Up workflow helps you review and refine what Alter remembers, and saved sources are clickable so you can jump straight back to where a memory came from.

A full guide is available in the documentation: Memory

Video:

New Features & Enhancements

• Scoped memory: First-class scoped memory with automatic capture, authorization hardening, and provenance tracking so Alter remembers context without you managing it manually.
• Memory Catch-Up: New workflow to review, refine, and humanize what Alter has learned.
• Memory sources: Open saved memory sources directly from settings; source links are clickable.
• Hub suggestion cards: Redesigned suggestion cards and faster initial Hub tab loading.
• Speech prompts: Speech prompts now always start a new chat.
• Running tool icons: Animated running tool icons in conversation view.
• Alter Docs: Embedded Help documentation updated with the latest features.

Bug Fixes & Stability

• Hub tabs: Left and right clicks on tabs work correctly on external screens (macOS 15).
• Computer use: Hardened URL handling; fixed background action workspace permissions.
• Edge Bar: Re-pin panels after display disconnect.
• Background Jobs: Custom actions with Skills and no workspace retains their ephemeral workspace to avoid permission issues.

Codex Sign-In & GPT 5.6

TL;DR: Sign in with Codex and use the new GPT 5.6 models. Computer use is dramatically faster — screen reads up to 77% quicker, and clicks/typing resolve in milliseconds instead of seconds on heavy apps like Notes and Spotify. Queue messages to steer your agent mid-task, and follow along with a cleaner grouped tool-call UI.

New Features & Enhancements

• Codex sign-in: New Codex provider integration with support for GPT 5.6 latest models
• Faster computer use: Major AppSense optimizations measured on real apps — AX screen reads up to 77% faster (Notes: 4.4s → 1.7s); element resolution for clicks and fills collapsed from 1–4+ seconds to ~1–4 ms on Spotify, Notes, and Reddit (~330–4100× faster). Plus improved app guidance and dynamic app-specific tools.
• Music app tools: Focused Spotify and Apple Music tools, plus a Deezer recipe.
• Queue messages: Send follow-up messages while the agent is still streaming to steer it mid-task.
• Grouped tool calls: Tool calls are now grouped in conversation view with cleaner summaries, attachments, and styling.
• Reasoning effort selector: Choose reasoning effort for supported models (Codex only, Alter next).
• Send shortcut: Configurable keyboard shortcut for sending messages.
• New Models: GPT 5.6 Luna & Terra

Bug Fixes & Stability

• Hub: Respects IME composition for Enter handling; fixed crash when opening QuickHub and text input undo crash.
• Ask user tool: Limited to 5 options per question.
• Audio: Fixed audio recording on input switch and a memory leak
• Search: Faster search index listing.
• Switcher: Ignores pasted switcher command tokens.

Background Computer Use

TL;DR: Until now, using AI to interact with your mac often meant pausing your own work. Not anymore. With Background Computer Use, Alter can interact with your machine and handle complex tasks in the background while you continue working on other things.

Make sure to join the Discord or Subscribe to our Youtube channel for a long form video tutorial.

New Features & Enhancements

• Background Computer Use: New tool that lets Alter operate apps in the background while you keep working — with a live preview card, dedicated agent browser with self-cleaning tabs, and optimizations for apps like Discord and Spotify.
• Ask Anything defaults: Computer use, Flow, and Ask User are now default tools for new sessions.
• Edge Bar: Inner-edge placement and updated reveal-distance logic for the activation hotspot.
• QuickHub: Now the default interface for new users; respects fullscreen position preference.
• Task sharing & export: Share task traces and export task chat as markdown.
• Transcripts: Duration shown on ready transcript task cards.
• AppSense: Re-enabled CEF and Qt access; waits for web areas before attaching.
• Scroll tool: Millisecond timing support.
• New models: GLM 5.2, Deepseek V4 Flash & Pro, Minimax M3, Cerebras Gemma 4 31B

Bug Fixes & Stability

• Translations: Added missing strings from recent releases.
• Onboarding: Restored missing grant button for automation permission.
• Edge Bar: Now displays over fullscreen apps.
• Markdown: Fixed currency rendering.