Limits & Quotas

Rate limits, quotas, and resource constraints for Go Bananas!.

Rate Limits

Default Limits

Resource	Limit	Window
API requests	60	Per minute
Image generations	30	Per minute
Uploads	20	Per minute
OAuth token endpoint	30	Per minute per IP

OAuth Rate Limiting

Rate limiting by endpoint type and IP identification

The /oauth/token endpoint has stricter rate limiting to prevent brute force attacks:

• 30 requests per minute per IP address
• Uses cf-connecting-ip header for IP identification
• Returns 429 Too Many Requests when exceeded

Admin API IP Detection

The Admin API uses only the cf-connecting-ip header for rate limiting to prevent X-Forwarded-For spoofing attacks. This ensures accurate rate limiting even when behind proxies.

Rate Limit Headers

Responses include rate limit information:

X-RateLimit-Limit: 60
X-RateLimit-Remaining: 45
X-RateLimit-Reset: 1705312800

Rate Limit Response

When exceeded:

HTTP/1.1 429 Too Many Requests
Retry-After: 30

{
  "error": "Rate limit exceeded. Try again in 30 seconds.",
  "retryAfter": 30
}

Custom Limits

Per-tenant limits can be configured:

curl -X PATCH /admin/tenants/{id} \
  -H "Authorization: Bearer <admin_session_token>" \
  -H "Content-Type: application/json" \
  -d '{"rate_limit_per_minute": 120}'

Storage Quotas

Default Quotas

Resource	Default	Max
Monthly storage	10 GB	Configurable
Images per session	1,000	None
Total images	Unlimited	Storage-based

Quota Tracking

Check usage:

curl https://gobananasai.com/api/usage \
  -H "X-API-Key: sk_live_xxx"

Response:

{
  "data": {
    "months": [
      {
        "month": "2024-01",
        "total_operations": 1523,
        "total_images": 2891,
        "total_bytes": 3019898880,
        "total_mb": 2880,
        "avg_duration_ms": 3421
      }
    ],
    "recent": { "bytes": 93323264, "images": 89 },
    "quota": {
      "monthlyQuotaMb": 10240,
      "monthlyUsedMb": 2880,
      "monthlyRemainingMb": 7360,
      "rateLimitPerMinute": 60
    }
  }
}

Quota Exceeded

When quota is exceeded:

HTTP/1.1 402 Payment Required

{
  "error": "Monthly storage quota exceeded",
  "code": "QUOTA_EXCEEDED"
}

Image Limits

Generation Limits

Parameter	Flash	Pro
Prompt length	8,192 chars	16,384 chars
Images per request	4	4
Character reference images	4	5
Min reference image size	512x512 px	512x512 px

File Limits

Parameter	Limit
Max upload size	20 MB
Minimum valid dimensions	1 × 1 px
Max total upload pixels	4,194,304 (about 4.2 MP; equivalent to 2048 × 2048)
Supported upload formats	PNG, JPEG, WebP, single-frame GIF

Wider images are allowed within the total-pixel limit. Resize larger uploads first; export animated GIFs as one frame. These are upload limits, not generation-resolution limits. The server validates actual encoded contents before saving.

Scene reference URL imports

Scene creation and updates allow at most 20 combined existing references and new URLs. URL imports have a 20 MiB aggregate encoded-byte limit per request, in addition to the per-image upload limits above. All URLs are fetched and validated before any image is saved; an invalid later image rejects the import without saving earlier ones. This prevalidation buffers at most the aggregate allowance and does not refetch URLs for storage.

Storage, quota or database failures after validation can still leave an earlier imported image saved. The import is not a transaction across all images; inspect retained images before retrying after such a failure. Split larger imports into smaller requests.

Storage Limits

Parameter	Limit
Thumbnail size	200 × 200 px
R2 object key	1024 chars
Metadata fields	64 KB total

Text Limits

Input Limits

Field	Min	Max
Prompt	1	16,384 chars
Negative prompt	0	1,024 chars
System instruction	0	512 chars
Character name	1	100 chars
Character base prompt	10	2,000 chars
Product name	1	100 chars
Style preset name	1	128 chars
Session ID	1	200 chars
Search query	1	200 chars

Description Limits

Field	Max
Character description	500 chars
Product description	500 chars
API key label	100 chars
Tags (each)	50 chars

Collection Limits

Array Limits

Collection	Max Items
Images per request	4
Character reference images (Flash)	4
Character reference images (Pro)	5
Characters per scene	5
Multi-char total ref images (Flash)	4
Multi-char total ref images (Pro)	5
Tags per character	20
Bulk delete IDs	20
Search results	200

Pagination Limits

Parameter	Default	Max
List limit	50	200
Search limit	15	50
History limit	50	200

API Key Limits

Per Tenant

Resource	Limit
API keys	10
Active keys	10

Key Format

Type	Format
Live key	sk_live_ + 32 chars
Test key	sk_test_ + 32 chars

Session Limits

Session Constraints

Parameter	Limit
Session ID length	200 chars
Images per session	1,000
Session lifetime	Indefinite

Database Limits

D1 Constraints

Parameter	Limit
Row size	1 MB
Query result	10 MB
Batch size	1,000 statements
Database size	Plan-dependent

Index Limits

Table	Indexes
images	5
characters	3
sessions	2

Cloudflare Limits

Workers

Resource	Free	Paid
Requests/day	100,000	10M+
CPU time	10ms	30s
Memory	128 MB	128 MB

D1

Resource	Free	Paid
Storage	5 GB	10+ GB
Reads/day	5M	25B
Writes/day	100K	50M

R2

Resource	Free	Paid
Storage	10 GB	Pay per use
Class A ops	1M/month	Pay per use
Class B ops	10M/month	Pay per use

KV

Resource	Free	Paid
Reads/day	100K	10M+
Writes/day	1K	1M+
Storage	1 GB	10+ GB

Provider Limits

Provider limits stack on top of the Go Bananas! tenant limits — your own quota is whichever the provider enforces, applied per tenant API key.

Gemini Rate Limits

Tier	RPM	TPM
Free	15	32K
Pay-as-you-go	1000	4M
Enterprise	Custom	Custom

RPM = Requests per minute, TPM = Tokens per minute. See Gemini API Pricing.

Gemini Content Limits

Parameter	Limit
Prompt	16,384 chars (Flash + Pro)
Reference images	14 (6 high-fidelity on Pro)
Output	PNG only

OpenAI gpt-image-2 Limits

Parameter	Limit
Prompt	32,000 chars
Reference images	16 (routed via /v1/images/edits)
Max edge	3,840px (both edges multiples of 16)
Aspect ratio cap	Long-to-short ≤3:1
Pixel range	655,360 – 8,294,400
Output formats	PNG, JPEG, WebP (with output_compression)
Quality tiers	low / medium / high / auto
Default request timeout	240s

OpenAI rate limits depend on your account tier and Organization Verification status; gpt-image-2 specifically requires Organization Verification. See OpenAI Rate Limits.

Increasing Limits

Tenant Limits

Contact admin to increase:

# Increase rate limit
curl -X PATCH /admin/tenants/{id} \
  -H "Authorization: Bearer <admin_session_token>" \
  -H "Content-Type: application/json" \
  -d '{"rate_limit_per_minute": 120}'

# Increase quota
curl -X PATCH /admin/tenants/{id} \
  -H "Authorization: Bearer <admin_session_token>" \
  -H "Content-Type: application/json" \
  -d '{"monthly_quota_mb": 20480}'

Cloudflare Limits

Upgrade to paid plans for higher limits:

• Workers Pricing
• D1 Pricing
• R2 Pricing

Provider Limits

• Gemini API Pricing — request quota increases in Google Cloud Console
• OpenAI Rate Limits — request increases via the OpenAI dashboard; gpt-image-2 requires Organization Verification

Next Steps

• Environment Variables →
• Troubleshooting →
• Scaling Guide →