One balance. Pay as you go. Better rates with Pro.
Standard rates apply to all GPU usage
Billed $239.90/year (4 months free)
20%+ off all GPU rates
Everything in Pay as You Go, plus:
| GPU Machine | Pay as You Go | Pro |
|---|---|---|
| Small: CPU For uploads & workflow setup (20 hr/mo) | Free | |
| Medium: T4, A4000 16 GB VRAM · 16 GB RAM · 8 vCPUs | $0.99/hr | $0.79/hr |
| Large: A10G, A5000 24 GB VRAM · 32 GB RAM · 8 vCPUs | $1.75/hr | $1.39/hr |
| X-Large: A6000 48 GB VRAM · 48 GB RAM · 28 vCPUs | $2.50/hr | $1.99/hr |
| X-Large Plus: L40S, L40 48 GB VRAM · 64 GB RAM · 28 vCPUs (API only) | $2.99/hr | $2.15/hr |
| 2X-Large: A100 80 GB VRAM · 96 GB RAM · 28 vCPUs | $4.99/hr | $3.99/hr |
| 2X-Large Plus: H100 80 GB VRAM · 180 GB RAM · 28 vCPUs | $7.49/hr | $5.99/hr |
| 3X-Large: H200 141 GB VRAM · 240 GB RAM · 24 vCPUs | $8.75/hr | $6.99/hr |
| 100% private workspace | ||
| Native ComfyUI (synced with official releases) | ||
| Upload custom models & install any nodes | ||
| Saved workflow environments | Up to 10 | Unlimited |
| Serverless API deployment | ||
| 200+ ready-to-run workflows | ||
| Storage | 10 GB | 200 GB |
| Persistent storage | Cleared after 90 days | |
| Support | Discord & Email | Priority support |
| Special rate for training runs | Pay as You Go | Pro |
|---|---|---|
| H100 (Training) | $4.49/hr | $3.59/hr |
| H200 (Training) | $5.25/hr | $4.19/hr |
| 8×H100 (Training) | $29.99/hr | $23.99/hr |
RunComfy uses one shared balance across Models (including Model APIs), ComfyUI (including API), and Trainer (including Trainer APIs).
Pro is optional and gives discounted GPU rates plus perks like extra storage and monthly credits.
RunComfy is the premier ComfyUI platform, offering ComfyUI online environment and services, along with ComfyUI workflows featuring stunning visuals. RunComfy also provides AI Models, enabling artists to harness the latest AI tools to create incredible art.
| Auto-release GPUs, no idle charges |
| Persistent checkpoints for resumable training |
| For deployed inference (Dedicated endpoints) | Pay as You Go | Pro |
| A6000 (Inference) | $2.50/hr | $1.99/hr |
| L40S/L40 (Inference) | $2.99/hr | $2.15/hr |
| A100 (Inference) | $4.99/hr | $3.99/hr |
| H100 (Inference) | $7.49/hr | $5.99/hr |
| H200 (Inference) | $8.75/hr | $6.99/hr |
| Support | Discord & Email | Priority support |
| Video Model | Price | Unit | Per $1 |
|---|---|---|---|
| Wan 2.5 | $0.045 | second | ~22 sec |
| Kling 2.5 Turbo | $0.07 | second | ~14 sec |
| Sora 2 | $0.10 | second | ~10 sec |
| Seedance 1.0 | $0.03 | second | ~33 sec |
| Image Model | Price | Unit | Per $1 |
|---|---|---|---|
| Seedream 4.0 | $0.03 | image | ~33 images |
| Flux Kontext Pro | $0.04 | image | ~25 images |
| Nanobanana | $0.04 | image |
These are sample rates. See each model's page for full pricing details.
RunComfy is great for classes, workshops, and tutorials because students can start ComfyUI instantly in the cloud (no local installs) and you get a consistent environment for everyone.
It also supports the full stack: use Models for instant image/video generation demos, and Trainer for LoRA / fine‑tuning on dedicated GPUs.
Billing is flexible: one shared balance, pay as you go, with optional Pro discounts on GPU usage. Learn more here.
You can get invoices for your balance top‑ups and subscriptions by following this guide:
Check out: Detailed Pricing Plans Q&A
| ~25 images |
| Qwen | $0.03 | image | ~33 images |
