Payment & Billing

Overview

MoltGhost uses usage-based billing for compute resources consumed by Agent Pods.

Pay only for what you use—billed per-second with no long-term commitments.

Deploy → Allocate GPU/Memory → Run → Pay per Second → Stop → Billing Pauses

Pricing Model

Per-Second Compute Billing across all resources:

Resource	Price (per hour)	Min Allocation	Notes
GPU
NVIDIA L40S	$0.80	1 GPU	7B-30B models
NVIDIA A100 80GB	$2.50	1 GPU	70B models
NVIDIA H100 80GB	$4.20	1-8 GPU	405B+ models
CPU/Memory
8 vCPU + 32GB	$0.15	2 vCPU	Lightweight agents
32 vCPU + 128GB	$0.60	8 vCPU	Heavy tooling
Storage
NVMe SSD	$0.15/GB	50GB	Model weights + data

Example Costs:

Llama 70B Agent (A100):
- Running 24h: $60/day
- Paused: $1.20/day (storage only)
- First 100s free trial

Billing Dimensions

Total Cost = Σ(Resource × Duration × Rate)

State	GPU/CPU	Memory	Storage	Networking
Running	✅ Billed	✅ Billed	✅ Billed	Per GB
Paused	❌ Idle	❌ Idle	✅ Billed	❌ Idle
Deploying	⏳ Billed	⏳ Billed	✅ Billed	❌ Idle
Terminated	❌ None	❌ None	❌ None	❌ None

Free Tier: First 100 seconds + 5GB storage across all agents.

Cost Management Dashboard

┌─────────────────────────────────────────────────────────────┐
│                      Billing Dashboard                       │
├─────────────────────┬─────────────────────┬─────────────────┤
│   Agent Usage       │     Cost Breakdown  │   Cost Control  │
├─────────────────────┼─────────────────────┼─────────────────┤
│ -  llama-agent     │ -  GPU: $45.20 (75%) │ [⏸️ Pause All] │
│   120h @ A100     │ -  Memory: $8.40     │ [🗑️ Delete Idle]│
│ -  qwen-bot        │ -  Storage: $2.10    │ [⚙️ Set Budget] │
│   80h @ L40S      │ -  Network: $1.30   │ [$50/mo Limit] │
│ Total: $78.45     │ Total: $57/mo      │                 │
└─────────────────────┴─────────────────────┴─────────────────┘

Live Metrics:

moltghost billing usage --live
# Agent: my-agent  Cost: $0.023/min  GPU: 85%  Est. Monthly: $33.12

Lifecycle Cost Control

graph TD
    A[Deploy] -->|GPU Active| B[Running: Full Cost]
    B --> C{Idle?}
    C -->|Yes| D[Stop: Storage Only]
    C -->|No| B
    D --> E[Resume: Full Cost]
    B --> F[Terminate: $0]
    
    style D fill:#90EE90
    style F fill:#FFB6C1

Cost-Saving Commands:

# Pause idle agents (saves 95% cost)
moltghost agent stop my-agent-qa

# Auto-pause after inactivity
moltghost agent set my-agent --auto-pause 15m

# Delete terminated pods
moltghost cleanup --orphans

Payment Flow

sequenceDiagram
    participant U as User
    participant P as Platform
    participant C as Compute
    
    U->>P: Add Funds / Credit Card
    U->>P: Deploy Agent
    P->>C: Provision Pod
    loop Per Second
        C->>P: Usage Metrics
        P->>U: Deduct Balance
    end
    U->>P: Stop/Terminate
    P->>C: Release Resources

Funding Options:

Credit Card (instant)
Bank Transfer (IDR/WIB, 1-2 days)
Crypto (USDC, instant)
Enterprise Invoicing

Minimum Balance: $5 (auto-recharge available)

Detailed Billing Example

70B Llama Agent, 1 month mixed usage:

Week 1: 24/7 Production     → 168h × $2.50 = $420
Week 2: 12h/day Work Hours → 84h × $2.50 = $210  
Week 3: Paused (storage)   → 0 GPU + $2.10 = $2.10
Week 4: Testing 4h/day     → 28h × $2.50 = $70
Network: 150GB × $0.10     → $15
───────────────────────────────────────
Total: $717.10/mo

Optimization: Auto-pause → $250/mo savings (65%)

Summary

Transparent, Predictable Compute Billing:

✅ Per-second granularity (no hourly minimums)
✅ Live cost dashboard + alerts
✅ Lifecycle controls save 50-90%
✅ Free trial seconds for testing
✅ Multi-currency + enterprise billing

Deploy with confidence—pause anytime, scale intelligently, pay precisely.

Next: Enterprise Features → SSO, VPC Peering, Dedicated Clusters

Pro Tip: Set --auto-pause 30m on dev agents to eliminate idle costs.