AetherAI v1.24.7

AetherAI 1.24.7

• Organize Settings into application preferences, AI models/providers and tools. Configure provider keys in a compact dialog; Enter saves, Escape and outside clicks dismiss it. Square switches disable providers without deleting encrypted keys.
• Reserve #B5F36D for enabled switches and thin answer underlines. Restore the neutral white default accent and retain other saved appearance choices.
• Improve model lists, selected-model checkmarks, rounded search fields, menu spacing and button feedback. Fix unreadable selected-model labels with black accents and scrollbar placement at rounded corners.
• Add code-block language headers, syntax highlighting and localized copy controls. Preserve code indentation and newlines when copying. Complete streamed answers without replaying the whole-message animation.
• Present sources as compact links with publisher favicon fallbacks. Offer one localized explanation prompt for complex math, programming and multi-step answers.
• Add dismissible notifications and update reminders in all ten interface languages. Installed Windows builds support explicit download and restart/install through GitHub releases. Older builds require one installer upgrade first; portable/macOS/Linux builds offer official download links.
• Publish Windows update manifests and block maps alongside installers, with asset-name and SHA512 checks before making a release public.
• Preserve the landing page's scroll-driven spiral, remove its Features arrow button and add a short “Scroll down...” hint on the left.

Windows installer upgrades retain the existing application identity and user data. No automatic installation occurs without a user action. macOS builds remain unsigned and unnotarized.

AetherAI v1.24.6

AetherAI 1.24.6

• Choose AetherAI, Claude, ChatGPT or DeepSeek visual preferences from a keyboard-accessible logo list in Customization. Logos retain their native brand colors; application colors, fonts, backgrounds, welcome text and the selected model remain unchanged.
• Add an animated Claude spark, ChatGPT pulse and DeepSeek dots. Make Claude motion clearly visible and retain the approved AetherAI fragment animation. Indicators respect reduced motion and pause in hidden windows.
• Keep the mascot out of the desktop interface and omit the cancelled CMD chat.
• Fix response controls overflowing narrow windows and complete Russian labels for providers, agents and the local profile.
• Show “Done” consistently for completed responses.

AetherAI v1.24.5

AetherAI 1.24.5

• Stream answers progressively and retain received text when generation is stopped.
• Place content-sized response controls alongside the model selector, with correctly anchored menus and a quota indicator below the row. Remove the Agents button and restore rectangular window controls.
• Fix composer focus, duplicate tooltips, code contrast in both themes, and Russian plugin and skill labels.
• Improve searchable model selection and show only known model details. Remove the response-check button and lengthy usage panels from model menus.
• Fix plugin menu clipping, keyboard navigation, and layouts in narrow windows.
• Limit website-specific clarification to website requests. Calculator interface requests offer desktop or browser options; games no longer open the website wizard.
• Keep image prompts out of visible replies and preserve requested media, numbers, and lettering. Add optional Pollinations cloud generation with an encrypted API key; without a key, use local SD Turbo. Cloud image requests use your provider balance.
• Use the default animated AetherAI logo for Discord Activity and remove custom image/GIF controls and the application-renaming button.

AetherAI v1.24.4

AetherAI 1.24.4

Composer suggestions now start immediately with a blinking terminal block cursor, without first showing the generic placeholder.

Suggestions use a shuffled queue that persists across app restarts: 140 examples each in English and Russian, and 36 each in the other supported languages. Questions do not repeat until the queue finishes, and adjacent repeats are prevented across cycles.

Each question remains visible for five seconds before erasing. Suggestions disappear while writing or composing text, and reduced motion shows a static example without a blinking cursor.

AetherAI v1.24.3

AetherAI 1.24.3

The composer now uses a single glass surface with consistent rounded edges, removing the color seam between the text area and controls.

Settings open in a centered, rounded window over a blurred workspace. The dialog fits smaller windows and keeps its contents scrollable.

The sign-in wordmark stays readable in the light theme.

AetherAI v1.24.2

AetherAI 1.24.2

The default interface is now light, with dark logos, solid white primary actions and frosted glass controls. Custom application colors remain available.

Connected providers and Ollama Cloud now show a single Disconnect action. The welcome screen has a flowing logo, and the composer types example questions, holds each for five seconds, then erases it. Suggestions pause while writing; reduced motion keeps the interface still.

Google sign-in uses your Google profile photo automatically. Your chosen avatar or initial takes priority, and unavailable photos fall back to your initial.

AetherAI v1.24.1

AetherAI 1.24.1

Mathematical replies now render fractions, roots, powers, sums, integrals and matrices offline. Formula rendering keeps code examples literal and blocks untrusted links, images and HTML commands.

The logo morphs from A into flowing waves and back on hover. Discord uses a new animated GIF URL to avoid the old cached image. Reduced motion keeps the app logo still.

AetherAI v1.24.0

AetherAI 1.24.0

MultiMind is now AetherAI. A new flowing A logo appears throughout the app, sign-in, website and desktop packages, with matching hover and thinking animations. Existing chats, models, presets and account data use their established storage identifiers so upgrades preserve them.

MultiMind v1.23.2

MultiMind 1.23.2

• Hide the profile avatar in the collapsed sidebar and close its menu when collapsing; prevent clicks while collapsed.
• Add spacing between background labels and the animation note.
• Apply saved appearance before showing the workspace, eliminating the default background and heading flash at startup.
• Correct cloud Vision capabilities for supported Groq/Gemini models and use provider-reported image modalities when available.
• Tell text-only models that image analysis is unavailable. Image attachments require an explicitly selected Vision model; no automatic switch to another provider.
• Preserve image MIME types in cloud requests and reject unsupported image requests before sending them.

Windows installer/portable, macOS Intel/Apple Silicon DMGs and Linux AppImage/deb are built and checked on native CI runners. Packages remain unsigned; macOS is not notarized.

MultiMind v1.23.1

MultiMind 1.23.1

• Rejected dangling symbolic links/junctions and checked file boundaries before creating parent directories.

• Added account sign-in with Google, email/password registration and six-digit email confirmation, plus a styled Google completion page.

• Added response Presets, an upward profile menu, account initials and custom avatars; presets require explicit consent for the selected model.

• Added SambaNova, model details and optional checks, color customization and selectable animated backgrounds.

• Hardened IPC, navigation, renderer CSP, external links, project file boundaries and account-session races.

• Updated Electron and dependencies; dependency audits reported zero known vulnerabilities. Packages use an explicit file allowlist and private files/templates are excluded from Git and installers.

• Added repeatable RLS setup and database audit SQL. The live app table is currently missing/not exposed; production sync requires applying and verifying the server policies.

• Updated six website download links to 1.23.1. Windows installer/portable, macOS Intel/Apple Silicon DMGs, Linux AppImage/deb are built by CI. Packages remain unsigned; macOS is not notarized.

Previous 1.21.0 changes:

• Added OpenRouter, Groq, Gemini and Cerebras connections with encrypted API key storage.
• Added Show all / Only free / Only paid model filters, persistent selection, provider icons, model pricing and reported quota counters.
• Added Local AI mode with independent local/cloud model selection and the local download catalog.
• Made ordinary conversation use a general-purpose assistant prompt; greetings and short messages no longer authorize file creation or start the specialist planner.
• Added readable model errors and separate handling for authentication, billing, request limits and provider overload.
• Fixed temporary OpenRouter in-flight spending holds incorrectly appearing as exhausted account quotas; HTTP-date Retry-After values are now honored.
• Documented verified OpenRouter limits and their API source; quota counters remain based on provider data.
• Updated the website and six download links to 1.21.0; packaged Windows installer/portable, macOS Intel/Apple Silicon DMGs and Linux AppImage/deb.
• Removed a tracked Python cache, excluded developer scripts and branding exports from desktop packages, and expanded syntax checking to every runtime JavaScript file.
• Optimized model filter persistence to save settings without rewriting the full chat history.

macOS packages are unsigned and not notarized. Built-in llama.cpp setup remains Windows-only; cloud providers and Ollama support the other desktop platforms.

Changes relative to v1.21.0: 9054 lines added, 1821 lines removed across 72 files. Binary files are marked with -.

File	Added lines	Removed lines
.github/workflows/cloudflare-check.yml	5	0
.github/workflows/desktop.yml	15	4
.gitignore	29	0
README.md	24	0
RELEASE_NOTES.md	13	1
SECURITY_AUDIT.md	11	0
account-memory.js	66	0
appearance.js	20	0
auth-config.example.json	4	0
background-effects.css	1	0
background-effects.js	4401	0
background-effects.js.LEGAL.txt	72	0
chat-intent.js	13	2
docs/_headers	3	1
docs/account-setup.md	57	0
docs/database-security-check.sql	12	0
docs/effects.css	1	1
docs/effects.js	256	9
docs/index.html	2	2
docs/memory-schema.sql	27	0
docs/source/README.md	2	0
docs/source/app-backgrounds.jsx	26	0
docs/source/backgrounds/PatternWaves.css	6	0
docs/source/backgrounds/PatternWaves.jsx	895	0
docs/source/backgrounds/PixelBlast.css	6	0
docs/source/backgrounds/PixelBlast.jsx	614	0
docs/source/backgrounds/SOURCES.md	11	0
docs/source/backgrounds/ShapeWaves.css	29	0
docs/source/backgrounds/ShapeWaves.jsx	848	0
docs/source/boot.js	1	1
docs/source/build-page.cjs	5	1
docs/source/build.cjs	1	0
docs/source/effects.css	3	0
docs/source/effects.jsx	25	5
docs/source/package-lock.json	201	1
docs/source/package.json	4	0
electron/account-auth.js	59	0
electron/account-config.js	5	0
electron/auth-result-page.js	9	0
electron/cloud-providers.js	9	6
electron/main.js	104	87
electron/ollama-cloud.js	3	2
electron/preload.js	15	1
electron/security.js	7	0
electron/user-memory.js	21	0
index.html	52	15
model-info.js	16	0
multimind.css	117	0
package-lock.json	480	1518
package.json	21	16
renderer.js	60	89
resources/background-previews/SOURCES.md	1	0
resources/background-previews/gateway.png	-	-
resources/background-previews/pattern.png	-	-
resources/background-previews/pixel.png	-	-
resources/model-icons/sambanova.svg	1	0
resources/reactbits-license.txt	25	0
scripts/security-check.cjs	8	0
styles.css	3	50
tests/account-memory.test.js	85	0
tests/appearance.test.js	8	0
tests/auth-email-diagnostics.test.js	20	0
tests/auth-result-page-smoke.cjs	6	0
tests/chat-intent.test.js	7	0
tests/model-mode-preload.cjs	12	0
tests/model-mode-smoke.cjs	130	5
tests/native-shell-smoke.cjs	2	1
tests/package-security.cjs	7	0
tests/sambanova.test.js	39	0
tests/security.test.js	7	0
tests/site-locales.test.js	3	1
tests/site-smoke.cjs	3	2

Full Changelog: v1.23.0...v1.23.1