Reference
Limits
Current platform limits for Capsul. Because Capsul is BYOK (bring your own key), the only hard limits are platform-side — AI usage limits are set by your provider.
Platform limits
| Limit | Value | Notes |
|---|---|---|
| Apps per user | 3 | Delete an app to free a slot |
| AI generations per hour | 60 | Rolling window, resets as old calls age out |
| Version history per app | 5 | Oldest removed when 6th is created |
| Generated app size | ~100 KB | Single-file HTML; typical apps are 20–60 KB |
AI usage limits (provider-side)
Each AI generation is a single API call. Typical output is 1,000–4,000 tokens per generation. Your provider's rate limits apply separately:
Groq
- •~14,400 requests/day (free)
- •6,000 req/min (free)
- •Varies by model
OpenRouter
- •Free models: 50 req/day
- •Paid models: credit-based
- •No daily cap on paid
NVIDIA NIM
- •Credit-based
- •Free credits on signup
- •No hard daily limit
Limits change frequently — check each provider's documentation for up-to-date numbers.
Self-hosted instances
If you self-host Capsul, all limits are configurable. The app limit and rate limits are enforced in lib/rate-limit.ts and the apps schema — you can adjust them freely in your own deployment.