Cost estimation — Arqel on Laravel Cloud

Disclaimer: the prices below are indicative and reflect the public Laravel Cloud pricing as of April 2026. Always confirm up-to-date values at cloud.laravel.com/pricing before committing to a budget. This page is part of the LCLOUD-005 case study.

Cost structure

An Arqel app in production pays for five categories:

1. Compute (web + workers + Reverb) — per instance hour.
2. Database (managed Postgres) — per monthly tier.
3. Cache/Queue/PubSub (managed Redis) — per monthly tier.
4. Egress (outbound bandwidth) — per GB.
5. Arqel-specific add-ons:
   • AI tokens (@arqel-dev/ai with OpenAI/Anthropic).
   • Reverb connections (included in compute, but affects sizing).
   • S3-compatible storage for FileField uploads.

Reference Laravel Cloud table (April 2026)

Resource	Tier	Approx. price/month
Web instance nano (256 MB)	—	$5
Web instance small (512 MB)	—	$12
Web instance medium (1 GB)	—	$24
Web instance large (2 GB)	—	$48
Postgres Hobby	256 MB / 1 GB storage	$0 (free tier)
Postgres Starter	1 GB / 10 GB	$20
Postgres Production	4 GB / 50 GB	$80
Postgres Scale	16 GB / 200 GB	$300
Redis Hobby	25 MB	$0
Redis Starter	256 MB	$15
Redis Production	1 GB	$50
Egress	—	$0.10/GB (after the first 100 GB free)

Add-ons outside Cloud:

Add-on	Approx. cost
OpenAI GPT-4o (input)	$2.50 / 1M tokens
OpenAI GPT-4o (output)	$10 / 1M tokens
Anthropic Claude Sonnet 4.7 (input)	$3 / 1M tokens
Anthropic Claude Sonnet 4.7 (output)	$15 / 1M tokens
S3-compatible storage (B2/R2)	~$5 / 100 GB

Calculator — three scenarios

Small — internal tool (≤10 users)

Item	Quantity	Monthly
Web nano × 1	24×7	$5
Worker nano × 1	24×7	$5
Reverb nano × 1	24×7	$5
Postgres Hobby	—	$0
Redis Hobby	—	$0
Egress (~5 GB)	—	$0
Cloud total		$15
AI tokens (rare, ~100k tokens)		$1
Total		~$16/month

Medium — B2B SaaS (10-50 users, occasional AI)

Item	Quantity	Monthly
Web small × 2	auto 1-3	$24-36
Worker small × 2	auto 1-3	$24-36
Reverb small × 1	24×7	$12
Postgres Starter	—	$20
Redis Starter	—	$15
Egress (~50 GB)	—	$0
Cloud total		~$95-120
AI tokens (~5M input + 1M output)		$22
Total		~$120-145/month

Large — public product (50-200 users, heavy AI)

Item	Quantity	Monthly
Web medium × 4	auto 4-8	$96-192
Worker medium × 4	auto 4-8	$96-192
AI worker medium × 2	auto 2-4	$48-96
Reverb medium × 2	sticky	$48
Postgres Production	—	$80
Redis Production	—	$50
Egress (~500 GB)	—	$40
Cloud total		~$460-700
AI tokens (~50M input + 15M output)		$350
S3 storage (200 GB)	—	$10
Total		~$820-1,060/month

Arqel-specific cost optimizations

1. Cache table queries. @arqel-dev/table uses cache()->remember() by default for getRecords() queries. Bump the TTL to 60s on low-volatility tables — drops 70% of Postgres queries.
2. AI prompt caching. Use Anthropic prompt caching (5 min TTL) to reduce 90% of input-token cost on long conversations. Already default in ClaudeProvider.
3. Exports on a low-priority queue. Run the exports queue on separate nano workers — cheap and does not compete with web for CPU.
4. Octane. Enable in cloud.yaml (octane: true). Reduces web-instance need by ~40%.
5. Postgres connection pooling with pgbouncer. Already default — do not disable.

Quick comparison with competitors

For the Medium scenario above (~$120/month on Laravel Cloud):

Platform	Equivalent cost	Setup effort
Laravel Cloud	$120	Zero (cloud:export)
Fly.io	~$95	Medium (write fly.toml)
Render	~$110	Low (Blueprint)
DigitalOcean App Platform	~$80	Medium (App Spec)
AWS App Runner + RDS	~$140	High (CDK or Terraform)
Heroku	~$200	Low

Details in comparison-other-hosts.md.

When it is not worth it

Laravel Cloud is expensive if:

• You have ≤3 users and everything fits on a $6 DigitalOcean droplet.
• Your app is stateless and write-heavy — Lambda + RDS comes out cheaper.
• You need regions outside the supported set (especially Africa/Oceania beyond ap-southeast).

In those cases, consider Fly.io or self-host on DigitalOcean with Forge.

Next steps

• Review per-component sizing → auto-scaling.md.
• Detailed comparison with other hosts → comparison-other-hosts.md.