Lab Notes/I Made Fusion and Qwen3.6 27B Build the Same Web App
LLM TestingJune 18, 2026Watch on YouTube

I Made Fusion and Qwen3.6 27B Build the Same Web App

I put OpenRouter Fusion and Qwen3.6 27B head-to-head and gave them the exact same prompt: build the same web app from scratch. Same goal. Same constraints. Same phased build. Very different results. In this video, I compare how a multi-model AI committee stacks up against a single 27B model when the task is actual software delivery, not just talking about code. The project was a real web app built in phases on fresh Linux VPSes, with each model responsible for turning the prompt into something usable. This wasn’t about benchmark scores or cherry-picked one-liners. I wanted to see which one could actually plan, build, adapt, and ship. For all prompts, code outputs, and info about the video, visit: https://tokenchaser.net Drop a comment with which models you want to see go head-to-head next.

Models2
Prompts3
Live HTML0
Files0

Video

Models Tested

Fusion
Quality
OpenRouter
Qwen3.6 27B
Q6_K_XL
RTX 3090ti & 3090

Prompts Used

1Phase 1
# Phase 1 — Core Player

You're on a fresh Ubuntu VPS. Install whatever dependencies you need. Build and deploy a web-based music player accessible on the local network at port 80.

## What to build

- **Drag & drop MP3 upload** — files go into a shared library, persisted to disk
- **Synced playback** — pause/play/seek on one device updates all connected browsers
- **Playlist manager** — create, rename, delete playlists, add/remove tracks
- **Crossfader** — smooth fade between tracks, adjustable 0–10s
- **Waveform scrubber** — click to seek, drag to set loop points
- **Frequency bar visualizer** — real-time audio visualization

## Tech

Your choice. Node.js or Python backend. WebSockets for sync. TypeScript frontend if you want. Whatever ships fast.

## Deploy

- Port 80, accessible at http://[VPS_IP]
- Auto-start on boot via systemd
- Persist uploaded files to disk

Build it. Deploy it. Make it work.
2Phase 2
# Phase 2 — Library & Control

Phase 1 is deployed and working. Extend it with these features.

## What to build

- **YouTube link support** — paste a URL, joins the queue, plays in sync
- **Speed/pitch control** — adjust playback speed without killing pitch
- **Request queue** — connected users browse library, vote on next track
- **Hot cues** — mark positions in a track, trigger instantly from the UI
- **Mixer panel** — overlay drums, beats, reverb, delay, EQ on the playing track. Adjustable volume per layer
- **3+ visualizer modes** — keep frequency bars, add circular spectrum and particle effects
- **Theme selector** — dark, neon, pastel

## Deploy

- Same port 80, same auto-start, same persistent storage
- Extend the existing codebase

Build it. Deploy it. Make it work.
3Phase 3
# Phase 3 — Polish & Intelligence

Phase 2 is deployed and working. Add the finishing touches.

## What to build

- **Auto-mix** — crossfade to next track when current one ends, BPM-aware beatmatching if possible
- **Scheduled playback** — set times of day for playlists to auto-start
- **AirPlay casting** — cast audio + visualizer to an AirPlay receiver on the network
- **Mood selector** — pick a mood (chill, party, focus, workout), auto-generates a playlist from the library based on tempo/energy
- **Multi-room support** — group devices into rooms, play different playlists in different rooms

## Deploy

- Same port 80, same auto-start, same persistent storage
- Extend the existing codebase

Build it. Deploy it. Make it work.