Freestyle

Agents 4.6/10

AI agent sandboxes by memory tier: XS (2GB) $0.0828/hr, S (4GB) $0.1656/hr, M (8GB) $0.3312/hr, L (16GB) $0.6624/hr, XL (32GB). Billed by wall time; effective ~$0.04032/vCPU-hr.

Freestyle scores 4.6/10 for free value in Agents. The main constraint: Paid — Memory-tier; XS (2GB) $0.0828/hr. AI agent sandboxes by memory tier: XS (2GB) $0.0828/hr, S (4GB) $0.1656/hr, M (8GB) $0.3312/hr, L (16GB) $0.6624/hr, XL (32GB). Billed by wall time; effective ~$0.04032/vCPU-hr.

Category Agents Kind Paid plan

Free value score

4.6 / 10

Show the math

Overall 4.6/10

value = 4.6/100

weighted_components_v1

Execution 1/10

value = 0.01 ×

fixed_log_v1

Connectivity 8.2/10

value = 8.2/100

policy_score_v1

Production controls 8/10

value = 8/100

policy_score_v1

Key metrics

1/mo Execution quota Uses stated quota when present; otherwise a conservative unpublished-quota floor.

1 Tool/model connectivity langchain

2 Production controls sandbox, sandboxes

$0.083/hr

2GB XS

4GB $0.17

What you get

Paid — Memory-tier; XS (2GB) $0.0828/hr. AI agent sandboxes by memory tier: XS (2GB) $0.0828/hr, S (4GB) $0.1656/hr, M (8GB) $0.3312/hr, L (16GB) $0.6624/hr, XL (32GB). Billed by wall time; effective ~$0.04032/vCPU-hr.

~$0.04/vCPU-hr wall-time rate undercuts Vercel Sandbox by ~3x for steady agent workloads

Gotchas

No gotchas documented yet. Contribute one if you know a constraint we missed.

Sources

"XS (2GB) at $0.0828/hr, S (4GB) at $0.1656/hr, M (8GB) at $0.3312/hr, L (16GB) at $0.6624/hr"
https://northflank.com/blog/ai-sandbox-pricing
"Freestyle vCPU (wall time) rate: $0.04032 per vCPU-hour"
https://www.reddit.com/r/LangChain/comments/1sr60mz/sandbox_pricing_calculator_vercel_vs_freestyle/

paid value agent-deployment

Open service ← All Agents

Alternatives

Same category · similar quota

CodeSandboxCodeSandbox · paid · Paid plan

4.6

Firecracker microVMs for AI coding agents. Pro: 160 hours/month of VM credits, on-demand VM credits at $0.15/hour, 1,000 new sandboxes/hour, 250 concurrent VMs. VM credit individua

1/mo12

Checked 26d ago

Paid onlyMetered

Fly.ioFly.io · paid · Paid plan

4.6

Persistent Linux microVMs (Sprites) that hibernate when idle and wake in 100–500ms. CPU $0.07/CPU-hr, memory $0.04375/GB-hr, hot NVMe storage $0.00068/GB-hr, 100GB per Sprite. Bill

1/mo11

Checked 26d ago

Paid onlyMetered

LangSmith DeploymentLangSmith Deployment · paid · Paid plan

4.6

Managed LangGraph agent hosting. Plus $39/seat/mo includes 1 free dev-sized deployment (unlimited runs), 10k traces/mo. Additional: $0.005/deployment run + $0.0036/min prod uptime

1/mo22

Checked 26d ago

Paid onlyPer-seatMetered