Features

Five tools. One interface.

Stop switching between apps. Chat, code, create images, generate videos, and search the web — all in one place, powered by three specialized AI models.

AI Chat

Conversations that actually go somewhere

Three specialized models — Taipei, Majuli, and Suzhou — each purpose-built for different workloads. Context that persists. Reasoning you can follow.

  • Extended Thinking for complex multi-step problems
  • 200K token context window on Taipei 3.1
  • Conversations organized automatically by topic
  • Markdown, code blocks, and LaTeX rendered natively
< 2s
avg. response time
How do I implement rate limiting in a Next.js API route?
T

Here's a clean approach using an in-memory store with sliding window...

const rateLimit = new Map()
Can you add Redis support?
Code Workspace

From idea to deployed app in minutes

Describe what you want to build. Tripplet writes the code, shows you a live preview, and deploys it with one click. No scaffolding. No context-switching.

  • Multi-file editor with syntax highlighting
  • Live browser preview updates as you chat
  • One-click deploy to a public URL
  • Supports React, HTML, Tailwind, and more
10×
faster prototyping
components/Hero.tsx
export default function Hero() {
return (
<section className="hero">
<h1>Build faster</h1>
</section>
)
}
Image Generation

Words become visuals in seconds

A full image pipeline with automatic prompt enhancement, style selection, and aspect ratio control. Go from idea to polished image without touching a design tool.

  • Prompt enhancement for higher quality output
  • Style presets: photorealistic, artistic, diagrammatic
  • Multiple aspect ratios (1:1, 16:9, 9:16, and more)
  • Regenerate with variations or refine with a follow-up message
89K+
images generated

Generated image

"A minimal desk setup with warm lighting, top-down view"
Video Generation

Animate your ideas, not your timeline

Transform text prompts into short videos up to 10 seconds. Cinematic styles, multiple aspect ratios, and motion presets that make outputs look intentional.

  • Durations from 2 to 10 seconds
  • Landscape, portrait, and square formats
  • Motion styles: smooth, dynamic, cinematic
  • Download MP4 or share directly from Tripplet
5s
avg. generation time
Cinematic16:95s
Extended Thinking

Watch the model reason before it responds

When you toggle Extended Thinking, Taipei 3.1 allocates a reasoning budget before forming its answer. Complex proofs, architecture decisions, and debugging become dramatically better.

  • 32K internal reasoning token budget
  • 23% better accuracy on complex reasoning tasks
  • Best for: math, code architecture, research synthesis
  • Off by default — toggle when you need depth
23%
accuracy improvement
Extended Thinking active

Let me break this down step by step...

First, consider the constraint that n must be positive...

The recurrence relation simplifies to...

The answer is O(n log n). Here's the complete proof...

See it in action

No setup. No tutorial required. Just open Tripplet and start building.