Everything in
one system.
Alpanzo AI bundles a comprehensive suite of AI capabilities — no plugins, no extensions. Chat, generate images, create 3D models, run code, speak, search, study, and build documents all from one interface.
Real-time Streaming Chat
Multi-turn conversations with full history context. Streaming responses with typing cursor, automatic retry, message pinning, per-message copy, edit, speak, pin, and continue actions.
AI Image Generation
Generate images from natural language prompts. Intelligent edit detection — modify, restyle, or add elements to previous images. Gallery storage, watermark, direct download.
3D Model Generation
Generate interactive 3D models from text. Three.js WebGL viewer with orbit controls, wireframe toggle, auto-rotate, and GLB download.
Study Mode
AI-powered flashcard decks, structured study notes, and multiple-choice quizzes — all with a built-in chat assistant. GitHub sync to keep your study data across devices.
Alpanzo Codex
Full Monaco code editor with 20 language support, AI code tutor chat, Pyodide Python runner, UI preview, code conversion between languages, debug and review tools.
Voice Conversation Mode
Full duplex voice interface powered by Alpanzo Deep. Tap the orb, speak naturally, receive spoken AI responses. 12-language support, conversation log, web search in voice.
Web Search Integration
Toggle real-time web search before any message. Results injected as context, keeping AI answers current. Also via /web prefix and inside Voice Mode.
Python Code Execution
Run Python directly in the browser via Pyodide (full Python 3 in WebAssembly). Real-time stdout/stderr output beneath each code block. Fully client-side and sandboxed.
Live Web Preview Pane
HTML, CSS, and JavaScript code blocks open in a split-pane live preview. Toggle rendered output vs source code. Export, open in new tab, save as project, or fullscreen.
Universal File Attachment
Attach PDFs, DOCX, XLSX, PPTX, CSV, code files, plain text, and images. Non-image files are text-extracted server-side. Drag-and-drop or paste from clipboard.
LaTeX Math Rendering
Full MathJax renders inline and display-mode LaTeX beautifully. Math symbol quick-insert menu. Supports AMS and mhchem packages for chemistry notation.
Document Generation
Generate and export PDF, PPTX, DOCX, XLSX, CSV, and TXT via /docs. AI crafts content, system converts — one button to download.
Three engines.
One interface.
Switch models from the top bar. Each engine is optimized for a different task class. All models support streaming, history context, Thinking Mode, and the full feature set.
Fast & Smart
The primary everyday model. Optimized for speed and general intelligence. Handles conversation, summarization, writing, and quick tasks with low latency.
- Fastest response times
- General conversation & writing
- Web search integration
- Study Mode & Codex chat
- Recommended for most users
Best Reasoning Model
Purpose-built for complex multi-step reasoning. Excels at mathematics, logical analysis, and tasks requiring chained inference across many steps.
- Advanced multi-step reasoning
- Mathematics & proof-writing
- Complex code architecture
- AI quiz & flashcard generation
- Thinking Mode enhanced
Complex Tasks & Voice
The deep model handles the longest, most nuanced tasks. Powers Voice Conversation Mode and Chat Migration. Designed for research and extended sessions.
- Long-form document analysis
- Powers Voice Mode
- Deep context retention
- Research & synthesis tasks
- Best for Study Note generation
Four ways
to interact.
The mode row above the input box switches between four modes. Each changes how Alpanzo interprets your message and which pipeline processes it.
/3d.Up and running
in four steps.
Create an Account
Visit the Sagittarius1 authentication portal to register for free. Your account syncs your name, username, avatar, and profession into Alpanzo's personalization engine automatically on sign-in.
Sagittarius1-auth.pages.devLaunch Alpanzo AI
Open the app at alpanzoai.pages.dev. Sign in with your Sagittarius1 credentials. Your session is stored locally in localStorage and re-validated with the server on every page load.
Choose a Model & Mode
Select your AI model from the top bar pill (Alpanzo, R8, or Deep). Choose between Chat, Image Gen, Vision, and 3D Gen modes. Optionally toggle web search via the 🌐 globe icon. Access Study Mode and Codex from the sidebar.
Install as PWA (Optional)
Install Alpanzo as a native app via your browser's install prompt or Add to Home Screen. Available on Android, iOS, and desktop Chrome or Edge. Works offline with cached assets via Service Worker.
Power-user
commands.
Type these commands at the start of any Chat mode message to trigger specialized workflows. All commands work in real-time with the current model.
Generate an AI image from a prompt. Edit intents like "change background", "make it anime", or "add fire" are auto-detected and intelligently merged with the last prompt using AI — the original subject is always preserved.
Generate a 3D model from a text description. Triggers the full Sagittarius1 3D pipeline — animated skeleton with cinematic progress bar, step indicators, and GLB export — then loads into the Three.js WebGL viewer.
Generate and export a complete document. Supported formats: pdf pptx docx xlsx csv txt. Syntax: /docs <type> <topic>
Plot an interactive mathematical function with Plotly. Supports sin, cos, tan, sqrt, log, ln, exp, abs, floor, ceil, round, pi, e, and exponentiation. Zoomable, pannable, hoverable. Download as PNG.
Search the web for images and display them as an interactive card grid (up to 8). Each card has full-screen preview, a download button, and a source link.
Search for videos. Results display as rich cards with thumbnail, duration, channel info, and platform badge. YouTube videos play inline in a modal.
Prefix any query to force a live web search regardless of the web toggle state. Results are injected as context before the AI answers.
Attach, analyze,
and work with files.
Three ways to attach: click + → Attach File, drag and drop onto the chat area, or paste from clipboard (Ctrl+V for images). Multiple files can be attached in one message.
Create, edit,
and remix images.
Switch to Image Gen mode
Click the 🎨 Image Gen button in the mode row, or use the /image command from Chat mode.
Write a descriptive prompt
Be specific about subject, style, lighting, and mood. Include style cues like "photorealistic", "anime", "watercolor", "3D render".
Edit with natural language
Send follow-up messages like "make it anime style", "change background to space", "add fire". Alpanzo detects edit intent and merges with the original prompt using AI.
Save or regenerate
Each image has 🔄 Again to regenerate the same prompt, and ⬇ Save to download as a watermarked PNG. Generated images are auto-saved to the Gallery tab.
Describe it.
See it in 3D.
Powered by the Sagittarius1 3D engine (Trellis). Type any description, watch the animated generation process live, then interact with a fully rotatable GLB model right inside the chat.
Switch to 3D Gen mode
Click the 📦 3D Gen button in the mode row, or type /3d followed by your description in Chat mode.
Watch the cinematic skeleton loader
An animated card appears — rotating 3D cube, pulsing grid cells, a vertical scan line, a live progress bar, and four labelled step indicators: Connecting → Structuring → Applying textures → Exporting.
Interact with the Three.js viewer
The GLB model loads into a sandboxed Three.js WebGL canvas with cinematic lighting, ACES filmic tone-mapping, a grid floor, and smooth OrbitControls. Drag to orbit, scroll to zoom, right-drag to pan.
Control and export
Controls: Reset (restore default camera), Wireframe (toggle mesh wireframe), Auto-rotate (continuous spin), and Download GLB to save for use in Blender, Unity, or Unreal.
Real-world
information.
Ground AI responses in current data by toggling real-time web search. Results are fetched, summarized, and injected as context before generation.
How to enable
- Click the 🌐 globe icon in the input row — it glows green when active.
- Use the
/webprefix for a one-time search without changing the global toggle. - In the Selection Popup, highlight any text and click Search.
Image & Video search commands
/wimg <query>— web image search, returns a grid of up to 8 image cards/vid <query>— video search, returns rich cards with inline YouTube player
LaTeX rendering
& live plots.
LaTeX Math Rendering
MathJax renders math automatically. Delimiters: inline $...$ or \(...\), display block $$...$$ or \[...\]. AMS environments and mhchem for chemistry.
Interactive Graph Plotting
Use /graph <expression> to render a live Plotly chart. Zoomable, pannable, hoverable. Click PNG to export. Click ⛶ to fullscreen.
Run Python.
Preview web code.
Python Runner (Pyodide)
Any Python code block has a ▶ Run button. Execution uses Pyodide — full CPython 3 compiled to WebAssembly. Everything runs in your browser, no server round-trip. stdout and stderr output renders beneath the block in a terminal-style box.
Web Preview (HTML / CSS / JS)
HTML, CSS, or JavaScript code blocks have a ▶ Preview button. Clicking combines all code blocks from that message and opens them in the Preview Pane on the right.
All code block features
- Language badge — auto-detected language label in the code header
- Copy button — copies full block source to clipboard
- Run / Preview button — executes Python or opens web preview
- Syntax highlighting — highlight.js with atom-one-dark theme
Learn smarter
with AI assistance.
Study Mode is a complete AI-powered learning environment built directly into Alpanzo. Open it from the sidebar or press Ctrl+Shift+S. Three tools — Flashcards, Study Notes, and Quizzes — all with an AI Study Assistant chat panel on the left.
Create decks manually or generate 5–50 cards from any topic with one click. Cards flip with a smooth 3D CSS animation. Navigate with ← → arrows, shuffle for random order, or reset to original sequence. The AI can generate a full deck with question/answer pairs from a topic description in seconds.
Full Markdown editor with live preview toggle, a rich formatting toolbar (Bold, Italic, H1–H3, lists, code, blockquote, divider), image attachments, drag-and-drop file import (TXT, MD, PDF), and one-click Markdown download. AI can fill or generate notes for any topic with structured headings, key concepts, examples, and a summary.
Generate 3–20 multiple-choice questions with four options and an explanation for each answer. Select your answer (A/B/C/D), then submit the full quiz to get your percentage score. Each wrong answer reveals the correct option in green and shows the explanation. Retake at any time. Score is saved to the sidebar.
Context-Aware AI Assistant
The AI chat panel always knows what you're looking at — current deck name, active flashcard (front and back), note content (up to 3000 chars), or current quiz questions. Ask for explanations, more examples, or related content without copy-pasting.
AI Flashcard Generation
Enter a deck title, topic description, and card count (5–50). The AI streams cards incrementally — you can start reviewing before generation even finishes. Each card is a clean front/back pair with smart question phrasing.
Apply AI Content to Notes
When the AI assistant generates content in the Study Notes tab, an Apply to Note button appears below the message. One click appends the AI content directly to your current note in Markdown format.
Apply Flashcards from Chat
If the AI generates a JSON flashcard array in the chat, an Apply as Flashcards button appears. Clicking it adds all cards to your active deck (or creates a new deck) instantly — no copy-pasting required.
Local Storage + GitHub Sync
All study data (decks, notes, quizzes) is saved to localStorage under alpanzo_study. Connect a GitHub repo in Settings to sync your study data across all your devices via the GitHub API — versioned and backed up automatically.
Full Mobile Support
Study Mode is fully responsive. On mobile: the sidebar slides in from the left, the AI chat panel slides in full-screen from the right via a floating AI button, and a bottom navigation bar switches between Cards, Notes, Quiz, and Library tabs.
GitHub Sync Setup
Click the GitHub button in the Study Mode sidebar footer. Enter your GitHub username, repo name, and a Personal Access Token with repo scope. Your data is saved to alpanzo-study/data.json in your repo and synced automatically after each change with a 3-second debounce. The same credentials are shared with Alpanzo Codex.
Keyboard shortcut
Press Ctrl+Shift+S from anywhere in Alpanzo to toggle Study Mode open or closed.
AI model selection in Study Mode
Study Mode uses whatever model you select in the top bar of its chat panel. For flashcard and quiz generation, Alpanzo R8 produces the most varied and accurate content. For note generation, Alpanzo Deep produces the most comprehensive structured notes.
Code, run, debug
with AI by your side.
Alpanzo Codex is a full-featured, browser-based code editor powered by Monaco (the engine behind VS Code) with an AI code tutor, Python execution, UI preview generation, multi-language conversion, and GitHub sync. Open from the sidebar or press Ctrl+Shift+C.
Supported Languages (20)
VS Code-Quality Editor
Full Monaco editor (the engine powering VS Code) with per-language syntax highlighting, IntelliSense, bracket matching, auto-indentation, and a custom codex-dark and codex-light theme. Custom color tokens: keywords in cyan, strings in amber, types in emerald, functions in blue. Auto-saves to localStorage on every change with a 800ms debounce.
Context-Aware AI Chat
The AI chat panel always injects the current file name, language, and up to 3000 chars of code as context. Ask for code review, explanation, optimization, tests, or documentation — all with your actual code already loaded. The AI references specific lines and patterns from your code. One-click Apply to Editor replaces your code with the AI's suggested version.
One-Click Code Actions
Six quick-action buttons appear above the chat input when a file is open: Review (bugs, best practices, security), Explain (step-by-step breakdown), Optimize (performance and readability), Tests (unit test generation), Docs (JSDoc/docstring generation), and Output (simulate execution via AI).
Generate Code from Description
Click AI Write in the editor toolbar, describe what you want in natural language, and AI generates the full code for your file's language. The generated code replaces the current file content. Supports all 20 languages. Uses the selected model (Alpanzo Deep recommended for complex code).
Multi-Language Code Conversion
Click Convert in the toolbar, select a target language from the full 20-language grid, and AI translates your entire file's code to the new language. The converted code is saved as a new file (e.g., main_python.py) with correct syntax and idioms for the target language.
Cross-Device GitHub Sync
Connect your GitHub repo (shared with Study Mode credentials) to sync all code files to alpanzo-study/codex-data.json. The sync button performs a full bidirectional merge — local files updated remotely (by another device) are pulled in, and local changes are pushed out. Conflict resolution uses updatedAt timestamps — newest wins.
Run Options
Click Run in the top bar (or press F5) to open the Run modal with three execution methods:
Debug Options
Click Debug in the top bar to open the Debug modal with two options:
- Debug with AI — AI performs a thorough static analysis and lists every bug, issue type, line number (where identifiable), description, and fix suggestion. Results stream into the Debug terminal tab. The full analysis is also sent to the AI chat as a message you can ask follow-up questions about.
- Debug with Pyodide — For Python files: actually executes the code via Pyodide and captures real runtime errors with full tracebacks in the Debug terminal tab. Each error line shows an Explain button that sends the error and your code to the AI tutor for a detailed explanation.
Terminal
The terminal panel slides up from the bottom with three tabs: Output (stdout from runs), Debug (debug analysis results), and Console (manual commands). The terminal input accepts commands — type clear to reset. Each error line has a hover-activated Explain button that sends the error to the AI chat with full code context.
Editor toolbar extras
- Format — AI cleans and formats your code without changing logic
- Copy — copies the entire file to clipboard
- Download — saves the file with the correct extension
- Theme toggle — switches between codex-dark and codex-light
- Word wrap — toggles long-line wrapping
- Zoom in/out — adjusts editor font size (10–22px)
Keyboard shortcut
Press Ctrl+Shift+C from anywhere in Alpanzo to toggle Codex open or closed. Inside the editor, Ctrl+S saves and F5 runs.
Full duplex voice
in every Indian language.
Powered by Alpanzo Deep via the Sarvam-M model (NVIDIA). Access via sidebar → Voice Conversation or press Ctrl+Shift+V. Speak naturally in any Indian language — get an AI response spoken back to you.
Supported Indian Languages
Analyze your screen
with precise reasoning.
Alpanzo can now read and interpret your screen with the new Screen Read AI mode. Capture your current view and let the AI extract data, provide the best detailed and precise reasoning, and answer complex visual questions instantly.
Real-time Screen Analysis
Instantly capture any part of your screen. Alpanzo breaks down UI elements, text, charts, and images, giving you an immediate understanding of whatever is in front of you.
Best-in-Class Reasoning
Goes beyond basic OCR. The AI uses precise logical reasoning to understand the context of your screen, making connections between different visual and textual elements.
Precise & Detailed Answers
Ask highly specific questions about your screen content. Get detailed, actionable answers whether you're debugging code, reviewing a complex spreadsheet, or reading a chart.
Code that lives
beyond the chat.
When AI generates substantial code, Alpanzo wraps it in an interactive Artifact card with preview, version history, and AI-powered editing.
Automatic Detection
Any code block of 14+ lines in HTML, React, SVG, Python, JavaScript, CSS, or Markdown is automatically converted into an Artifact card with a live preview thumbnail. Detection is based on language tag and content patterns.
Visual Preview
HTML and React artifacts render inside an iframe thumbnail directly in chat. Click to open in the full preview pane. SVG artifacts preview against a neutral background. All previews are sandboxed via srcdoc.
AI-Powered Editing
Type natural language instructions in the edit bar — "make the button red", "add dark mode", "animate the title" — and AI applies the change while preserving overall code structure.
Version History
Every AI edit creates a new version. Use Undo and Redo buttons to travel through up to 20 versions. The version counter shows your current position in the stack.
Export & Download
Download any artifact as a file with the correct extension (.html, .jsx, .py, .js, etc.). Copy the full source code to clipboard. Save to the Projects sidebar for access across sessions.
Supported Types
HTML pages, React components, SVG graphics, Python scripts, JavaScript modules, CSS stylesheets, and Markdown documents. Type is detected automatically from language tag and content patterns.
Live side-by-side
rendering.
Save and reopen
code artifacts.
Saving a project
- In the Preview Pane, click Save — enter a project name.
- On any AI message, click the 📁 Save to Project action button.
Managing projects
- Click a project name to reopen its code in the Preview Pane.
- Hover to reveal ✏ Rename and 🗑 Delete icon buttons.
- Use the search bar at the top of the Projects panel to filter by name.
Continue anywhere,
anytime.
Since chats are local-only, migration lets you carry any conversation to another browser or device using an AI-generated contextual summary.
Built around
you.
Open Settings from the ⚙️ icon in the sidebar or press Ctrl+,. Use the search bar at the top of Settings to find any option instantly.
8 Color Themes
Default, Midnight, Forest, Rosé, Ocean, Sunset, Neon, Slate. Applied instantly via CSS variables.
Font Size & Compact Mode
Adjust message font size from 12px to 18px. Compact mode reduces the gap between messages.
Sound, Scroll & Timestamps
Toggle sound effects, auto-scroll to follow streaming output, and show/hide timestamps on all messages.
Thinking Mode
Enable to show AI's internal reasoning as a collapsible <think> block with pulse animation while streaming.
Enter to Send
Toggle between Enter to send (default) and Ctrl+Enter to send. Shift+Enter always creates a new line.
Nickname, Profession & Hobby
Injected into the AI's system context so responses are tailored to you. Auto-populated from your Sagittarius1 account.
AI Behavior Prompt
Define a custom system prompt to shape the AI's personality, tone, and style for all conversations.
Editable Keyboard Shortcuts
All keyboard shortcuts are fully rebindable. Click Edit next to any shortcut, press your desired key combination.
Local-first Storage
All chat history, gallery images, projects, Study Mode data, and settings live only in your browser's localStorage.
Move at the
speed of thought.
All shortcuts are fully customizable in Settings → Shortcuts.
Your data.
Only yours.
Alpanzo AI was built with a privacy-first philosophy. We never save your conversations, study data, or code to any server, database, or cloud storage. Everything lives exclusively in your browser.
Chat history · Image gallery · Saved projects · All settings · Study Mode data (decks, notes, quizzes) · Codex files · GitHub credentials (in session data) · Sagittarius1 session token
Clearing browser cache · Switching browsers · Different device · Private/incognito mode · Uninstalling the PWA · Settings → Data → Clear. Use GitHub Sync to protect Study Mode and Codex data.