Infrastructure

Cloudflare Workers Cost Estimator

Estimate your monthly Workers, KV, D1, R2, and Durable Objects bill based on real-world usage — before you scale.

Workers Invocations

Free: 100K req/day. Standard: $0.30 per 1M requests after first 10M/month

Avg requests per day
Per invocation; Free ≤ 10ms

Workers KV

Free: 100K reads/day, 1K writes/day. $0.50/M reads, $5/M writes beyond

$0.50/GB-month after 1 GB free

D1 SQLite Database

Free: 5M rows read/day, 100K rows written/day. $0.001/M rows read, $1/M rows written

$0.75/GB-month after 5 GB free

R2 Object Storage

Free egress always. $0.015/GB stored, $4.50/M Class A ops, $0.36/M Class B ops

10 GB free
PUTs, POSTs — 1M free
GETs — 10M free

Durable Objects

$0.15/million requests + $12.50/million GB-seconds compute

$0.20/GB-month
Estimated Monthly Cost
$0.00
Across all configured products
Workers $0.00
KV Storage & Ops $0.00
D1 Database $0.00
R2 Storage & Ops $0.00
Durable Objects $0.00
Free
$0
100K req/day
Standard
$5/mo
+ your usage
Enterprise
Custom
Volume discounts

How Cloudflare Workers pricing works

Workers uses a consumption-based model with generous free allowances. The Free plan gives you 100,000 requests per day and up to 10ms CPU time per invocation — enough for many hobby projects and low-traffic APIs.

The Standard plan ($5/month) removes the daily cap and allows up to 30 seconds of CPU time. You pay $0.30 per additional million requests beyond the first 10M included, and $0.02 per million GB-seconds of CPU.

The key insight: Workers costs are almost always dominated by volume, not duration. A typical API returning JSON at 5ms CPU time costs roughly $0.30 per million requests — scaling to 100M requests/month costs ~$27 in request fees plus minimal CPU charges. Compare this to Lambda ($0.20/M invocations) or Cloud Run (more complex pricing) and Workers often wins for edge-latency-sensitive workloads.

KV vs D1 vs Durable Objects

KV is a globally replicated key-value store — reads are fast everywhere, writes propagate eventually (seconds to minutes). Best for config, feature flags, and cached data.

D1 is SQLite at the edge, with consistent reads from a single region and replicas elsewhere. Best for structured relational data, user records, and content.

Durable Objects offer strongly consistent, stateful compute — one object instance, one location, real coordination. Best for collaborative features, WebSocket chat, and rate limiters.