Quickstart

Get started with AI on CLORE.AI in 5 minutes.

New to CLORE.AI? This guide will help you run your first AI workload.

Step 1: Create Account

1. Go to clore.ai

2. Click Sign Up

3. Verify your email

Step 2: Add Funds

1. Go to Account → Deposit

2. Choose payment method:

   • CLORE - Native token (often discounted)

   • BTC - Bitcoin

   • USDT/USDC - Stablecoins

3. Deposit minimum $5-10 to start

Step 3: Find a GPU

1. Go to Marketplace

2. Use filters:

   • GPU Type: RTX 3090, RTX 4090, A100

   • VRAM: 12GB+, 24GB+, 40GB+

   • Price: Sort by lowest

Important: Always check RAM and network speed, not just GPU!

• RAM: 16GB+ minimum for most AI workloads (vLLM, ComfyUI, SD WebUI)

• Network: 500Mbps+ recommended for faster model downloads

• Low RAM/network = long startup times or failures

Server Requirements Quick Reference

Workload	Min RAM	Min VRAM	Min Network
Ollama (7B)	16GB	8GB	200Mbps
vLLM (7-8B)	16GB	16GB	500Mbps
Stable Diffusion	16GB	8GB	500Mbps
ComfyUI (SDXL)	16GB	8GB	500Mbps
ComfyUI (FLUX)	32GB	16GB	1Gbps

GPU Quick Reference

Task	Minimum GPU	Recommended
Chat with AI (7B models)	RTX 3060 12GB	RTX 3090 24GB
Image Generation (SDXL)	RTX 3070 8GB	RTX 3090 24GB
Image Generation (FLUX)	RTX 3090 24GB	RTX 4090 24GB
Video Generation	RTX 4090 24GB	A100 40GB
Large Models (70B+)	A100 40GB	A100 80GB

Step 4: Create Order

1. Click Rent on your chosen GPU

2. Choose order type:

   • On-Demand - Fixed price, guaranteed

   • Spot - Bid price, 30-50% cheaper

3. Configure:

   • Docker Image: See below for your task

   • Ports: 22/tcp (SSH) + app-specific ports

   • Command: Startup script

Quick Deploy Configs

Chat with AI (Ollama)

Image: ollama/ollama
Ports: 22/tcp, 11434/http
Command: ollama serve & sleep 5 && ollama pull llama3.2

Image Generation (ComfyUI)

Image: yanwk/comfyui-boot:cu126-slim
Ports: 22/tcp, 8188/http
Environment: CLI_ARGS=--listen 0.0.0.0

Image Generation (Stable Diffusion WebUI)

Image: universonic/stable-diffusion-webui:latest
Ports: 22/tcp, 8080/http

High-Performance LLM (vLLM)

Image: vllm/vllm-openai:latest
Ports: 22/tcp, 8000/http
Command: vllm serve mistralai/Mistral-7B-Instruct-v0.2 --host 0.0.0.0 --port 8000

Note: Requires 16GB+ RAM. Gated models (Llama) need HUGGING_FACE_HUB_TOKEN env var.

Step 5: Connect

After order starts:

1. Go to My Orders

2. Find connection details:

   • SSH: ssh -p <port> root@<proxy-address>

   • Web UI: Use the http_pub URL (e.g., https://xxx.clorecloud.net)

Startup Time: First launch downloads models and dependencies. This can take:

• Ollama: 2-5 minutes

• vLLM: 5-15 minutes (downloading model from HuggingFace)

• SD WebUI/ComfyUI: 10-20 minutes

HTTP 502 errors during this time are normal. Wait for startup to complete.

Verify Your Service

After deployment, check if your service is ready:

# Check Ollama
curl https://your-http-pub.clorecloud.net/api/tags

# Check vLLM
curl https://your-http-pub.clorecloud.net/health

# Check SD WebUI / ComfyUI
# Simply open the http_pub URL in browser

What Do You Want to Do?

Chat with AI

Start here:

• Ollama - Easiest setup

• Open WebUI - ChatGPT-like interface

Advanced:

• vLLM - Production serving

• DeepSeek-V3 - Best reasoning

Generate Images

Start here:

• Stable Diffusion WebUI - Full features

• ComfyUI - Node workflows

Advanced:

• FLUX.1 - Best quality

• SDXL Turbo - Real-time

Generate Videos

Start here:

• Stable Video Diffusion - Image to video

Advanced:

• Hunyuan Video - Text to video

• Wan2.1 - High quality T2V

Work with Audio

• Whisper - Transcription

• Bark TTS - Text to speech

• Stable Audio - Music generation

Analyze Images

• Llama 3.2 Vision - Image understanding

• LLaVA - Visual QA

Train Models

• Jupyter - General ML

• DreamBooth - Custom image models

• Fine-tune LLM - Custom language models

Pricing Overview

Typical CLORE.AI marketplace rates:

GPU	CLORE/day	~USD/hour	Best For
RTX 3060 12GB	~80	$0.02	Budget, SD 1.5
RTX 3090 24GB	~150	$0.03	SDXL, 7B-13B LLMs
RTX 4090 24GB	~200	$0.04	FLUX, fast inference
A100 40GB	~400	$0.08	Large models, training
A100 80GB	~600	$0.12	70B+ models, video

Prices vary by provider. Spot orders can be 30-50% cheaper.

Tips for Beginners

1. Check RAM first - 16GB+ RAM is critical for AI workloads

2. Check network speed - 500Mbps+ for faster startup

3. Start small - Use On-Demand orders for testing (Spot can be interrupted)

4. Use ready images - Pre-built Docker images save setup time

5. Wait for startup - 502 errors are normal during first 5-20 minutes

6. Save your work - Download outputs before order ends

7. Use CLORE tokens - Often get better rates

Troubleshooting

HTTP 502 for a long time

1. Check RAM - Server must have 16GB+ RAM

2. Check VRAM - Must match model requirements

3. Wait for download - First run downloads models (5-20 min)

4. Check network - Slow network = longer downloads

Service won't start

1. Low RAM - Most common issue. Upgrade to 16GB+ RAM server

2. Low VRAM - Model doesn't fit. Use smaller model or larger GPU

3. Wrong port type - Use http (not tcp) for web interfaces

Need Help?

• Troubleshooting

• GPU Comparison

• Discord

• Telegram

Next Steps

Once comfortable:

1. Try more complex workflows

2. Explore different models

3. Set up automated pipelines

4. Consider on-demand for production