Token Chaser
AI · Homelabs · Benchmarks
Lab Notes/Qwen3.6 27B Q8 vs Claude Opus 4.6 | Head-to-Head
LLM TestingMay 28, 2026Watch on YouTube

Qwen3.6 27B Q8 vs Claude Opus 4.6 | Head-to-Head

In this video, I put Qwen3.6 27B Q8XL Unsloth head-to-head against Claude Opus 4.6 to see how a local model running on consumer GPUs compares against one of the strongest frontier coding models. Both models were given the same prompts at the same time so we could compare speed, coding ability, creativity, interactivity, UI polish, and how well each one handled more advanced single-file HTML projects. In this test, the models build: • Smart Home Energy Command Center • Agent Automation Studio • Tiny Planet Terraformer Game This is not meant to be a perfect scientific benchmark. It is more of a real-world coding head-to-head to see how these models actually behave when given the same tasks side by side. Let me know which model you think won and what matchup I should test next. #Qwen #Qwen36 #Claude #ClaudeOpus #LocalAI #LLM #AICoding #OpenSourceAI #RTX3090 #Unsloth #TokenChaser

Models2
Prompts3
Live HTML6
Files6

Video

Models Tested

Qwen3.6 27B
Q8_K_XL
RTX 3090ti & 3090
Claude Opus
Default
Anthropic

Prompts Used

1Smart Home Energy Command Center
You are in a coding head-to-head competition against another AI model. Your goal is to create the best possible result from this prompt.

Create a single HTML file for a polished Smart Home Energy Command Center dashboard.

This should feel like a real production web app for monitoring and controlling a modern smart home, not a basic analytics mockup. The dashboard should include live energy usage, solar production, battery storage, room-by-room power draw, smart device activity, cost savings, weather impact, and automation suggestions.

Include realistic sample data for a home with solar panels, battery backup, EV charging, smart HVAC, lighting, appliances, security cameras, and multiple rooms.

The dashboard should include:
- A hero summary section showing current power usage, solar production, battery percentage, estimated daily cost, and grid import/export status
- A room-by-room energy map showing which areas are using the most power
- A solar and battery flow visualization showing energy moving between solar, battery, home, EV charger, and the grid
- Smart device cards with status, usage, and toggle controls
- An EV charging panel with charge level, charging speed, estimated finish time, and cost
- Weather impact cards showing temperature, sunlight, and forecast effect on energy usage
- A cost savings chart comparing solar usage, battery usage, and grid usage
- An automation recommendations section with actions like “delay EV charging,” “pre-cool the house,” or “run dishwasher during solar peak”
- A searchable device usage table

Add interactive controls so the user can toggle devices on and off, switch between Today, Week, and Month views, filter devices by room, simulate cloudy weather, and mark automation suggestions as applied.

Make the UI premium, modern, and visually exciting with a dark glassmorphism tech-style design, smooth animations, glowing accents, clean spacing, responsive layout, hover effects, and strong visual hierarchy.

Keep everything in one HTML file with the CSS and JavaScript included. Name the file smart-home-energy-command-center.html.

Once you are finished, go back over all of your code, test the logic mentally, make sure everything works correctly, and fix anything that may be broken or incomplete before giving the final answer.
2Agent Automation Studio
You are in a coding head-to-head competition against another AI model. Your goal is to create the best possible result from this prompt.

Create a single HTML file for an AI workflow builder called Agent Automation Studio.

This should be more than a simple mockup. The user should be able to build an automation workflow visually using draggable nodes on a canvas, connect the nodes together, edit node settings, and preview the automation flow.

The app should include a sample workflow already loaded for a YouTube AI model testing pipeline. Example flow: New Video Idea → Generate Prompt Set → Run Local Model Test → Run Frontier Model Test → Compare Outputs → Create Thumbnail Notes → Publish Results to TokenChaser.net → Draft YouTube Description.

Include these node types:
- Trigger
- AI Prompt
- Local Model Run
- API Model Run
- Data Parser
- Condition
- Human Review
- Website Publish
- YouTube Draft
- Notification

The user should be able to:
- Add new nodes from a sidebar
- Drag nodes around the canvas
- Select a node and edit its title, type, model, status, and notes in a settings panel
- Delete selected nodes
- See connection lines between nodes
- Toggle nodes between Pending, Running, Complete, and Failed
- View a live workflow summary that updates based on the nodes
- Export the workflow as JSON in a visible output panel

Make the interface feel like a real SaaS workflow builder, with a left toolbox, center canvas, right settings panel, top status bar, and bottom run log. Use polished styling, smooth interactions, responsive layout, clear visual hierarchy, and useful sample data.

Keep everything in one HTML file with the CSS and JavaScript included. Name the file agent-automation-studio.html.

Once you are finished, go back over all of your code, test the logic mentally, make sure everything works correctly, and fix anything that may be broken or incomplete before giving the final answer.
3Tiny Planet Terraformer Game
You are in a coding head-to-head competition against another AI model. Your goal is to create the best possible result from this prompt.

Create a single HTML file for a fun idle/clicker strategy game called Tiny Planet Terraformer.

The player starts with a small barren planet floating in space and must terraform it by generating energy, water, plants, atmosphere, and life. The player can click the planet to generate resources, buy upgrades, unlock new biomes, trigger weather events, and watch the planet visually evolve over time.

The game should include:
- A start screen
- A rotating tiny planet in the center
- Click-to-generate resource mechanics
- Resources for energy, water, oxygen, plants, and coins
- Upgrade buttons with increasing costs
- Unlockable stages like Barren Rock, Blue World, Green World, Forest Planet, and Living Planet
- Random events like meteor showers, solar boosts, rain storms, and alien visitors
- Animated stars, particles, clouds, rain, lightning, and orbiting satellites
- A progress meter showing terraforming percentage
- Win condition when the planet reaches 100% terraformed
- Restart button and high score or best time stored locally

Make the game feel polished, colorful, relaxing, and fun. The planet should visibly change as the player progresses, with new colors, plants, clouds, water, and life appearing. It should be playable, interactive, and satisfying, not just a visual demo.

Keep everything in one HTML file with the CSS and JavaScript included. Name the file tiny-planet-terraformer.html.

Once you are finished, go back over all of your code, test the logic mentally, make sure everything works correctly, and fix anything that may be broken or incomplete before giving the final answer.

Live HTML Outputs

6
Qwen3.6 27B
Q8_K_XLRTX 3090ti & 3090
3
HTML
agent-automation-studio-qwen.html
38.6 KB
OpenDownload
HTML
smart-home-energy-command-center-qwen.html
46.4 KB
OpenDownload
HTML
tiny-planet-terraformer-qwen.html
34.0 KB
OpenDownload
Claude Opus
DefaultAnthropic
3
HTML
agent-automation-studio-claude.html
41.2 KB
OpenDownload
HTML
smart-home-energy-command-center-claude.html
47.7 KB
OpenDownload
HTML
tiny-planet-terraformer-claude.html
40.1 KB
OpenDownload