Verdent uses a credit-based system to manage computational resources. Understanding credit usage and monitoring helps you optimize costs effectively.

Credit Monitoring

Web Dashboard

Monitor your credit balance through User Menu → Dashboard or visit verdent.ai/dashboard:
  • Account name, current plan, and credit balance
  • Total credits (subscription + top-ups) and remaining credits
  • Detailed token consumption across sessions
  • Usage patterns and trends
  • Model selection impact
  • Cost optimization opportunities

Subscription Plans & Quotas

PlanPriceCredits/MonthBest For
Free TrialFree100 (7 days)New users
Starter$19/month320Light users
Pro$59/month1,000Regular users
Max$179/month3,000Heavy users
Credit System:
  • Credits refresh monthly per billing cycle
  • Consumption varies by model type
  • Frontier models consume more credits
Flexible Top-ups: 240 (20),720(20), 720 (60), 960 (80),1,200(80), 1,200 (100), or 2,400 credits ($200)
Credits are shared across Verdent VS Code and Desktop, enabling seamless workflow between platforms.

Parallel Agent Usage

Each workspace with an active task consumes credits independently:
Active WorkspacesCredit Usage
1 workspace1x baseline
2 workspaces2x baseline
3 workspaces3x baseline
To optimize parallel usage:
  • Complete and rebase workspaces promptly
  • Delete workspaces when done
  • Idle workspaces (no active conversation) don’t consume credits

Managing Long-Running Tasks

Quick Tips:
  • Use Plan Mode to understand scope before execution
  • Start fresh tasks between major work to reset context
  • Complete atomic work units (feature → test → commit)
  • Watch for slower responses as sessions grow longer

FAQs

Common causes:
  • Multiple parallel workspaces - Each active workspace uses credits independently
  • Large context - Long conversations use more tokens
  • Frontier models - Higher cost than efficient models
Check your active workspace count in All Workspaces.
No. Workspaces with no active conversation don’t consume credits. Credits are only used when making AI requests.

See Also