Runloop - Pricing

Try Our Pro Plan for Free!

All Pro Features*

No Credit Card Required

$50 Usage Credit

Start Your Free Trial

Trial sandbox limits (lifted on upgrade):

3 Running Devboxes

5 blueprints

10 snapshots

3 objects

Choose your plan

All plans include devbox sandboxes, public benchmarks, and usage-based compute billing.

Basic

Radar Features

Free

+ usage based-compute

100 GB Free Storage

1st month FREE

Get Started Free

Access to Public Benchmarks

Devboxes and Blueprints

Snapshots

Email Support

Pro

Recommended

$250

/month + usage

For teams running agents in production

1 TB Free Storage

Start Free Pro Trial

Everything in Basic plan, plus:

Suspend and Resume

Repo Connections

Custom Benchmarks

Access to Beta Features

Slack Support

Enterprise

Radar Features

For regulated industries and high-volume workloads

Talk to Sales

Everything in Pro plan, plus:

Reinforcment Fine-Tuning

Priority Support

Deploy to VPC

Custom Free Storage

Faq's

Everything You Need to Know

We’re dedicated to solving the complex challenges of productionizing AI for software engineering at scale.

Pricing and Billing

Platform and Infraestructure

Getting Started

How does the free trial convert to a paid plan?

New accounts start with $50 in usage credits and full Pro feature access. No credit card is required. Once the trial usage limit is reached, you’ll be prompted to choose how to continue: switch to the Basic plan (free subscription with usage billed based on what you use), or upgrade to a paid Pro or Enterprise plan. There is no automatic charge when the trial ends—you decide how to proceed.

What does a typical agentic workload cost on the Pro plan?

A standard 2-CPU / 4 GB devbox running 8 active hours per day costs approximately $2.73/day in compute (2 x $0.108 + 4 x $0.0252, multiplied by 8 hours). At 22 working days per month that is roughly $60 in compute plus the $250 Pro base. Teams running parallel benchmark jobs or high-concurrency evaluations should contact us for volume estimates tailored to their workload profile.

Do I get charged for suspended devboxes?

No. Suspended devboxes do not accrue compute or memory charges. Billing for CPU and RAM stops the moment suspension begins. Storage charges continue while a devbox is suspended, but compute billing stops entirely. This makes Runloop significantly more cost-efficient for agentic workloads with variable or intermittent load -- you pay only for active execution.

Why do AI coding agents need different infrastructure than standard serverless?

Traditional serverless platforms optimize for stateless, short-lived function execution. AI agents are long-running, stateful, and bursty -- they need persistent filesystems, fast snapshot and resume cycles, and the ability to suspend and restart without data loss. Runloop devboxes are full development environments with snapshot branching, browser support, and suspend-resume designed specifically for agent execution patterns and the economics of workloads that spend significant time idle between active runs.

What makes Runloop's infrastructure enterprise-grade?

Runloop provides microVM-level hardware isolation between tenants, SOC2 Type II certification, HIPAA and GDPR compliance, Credential Gateway for opaque token injection (agents never see raw secrets), network egress policies per devbox, and VPC deployment for teams that need infrastructure inside their own cloud account. Enterprise plans include priority support, custom SLA commitments, and dedicated onboarding.

Does the Enterprise plan include SOC2 and compliance certifications?

Yes. Runloop is SOC2 Type II certified and HIPAA and GDPR compliant. VPC deployment runs Runloop infrastructure inside your own cloud account with customer-managed network policies and data residency controls. Compliance documentation and audit reports are available via the Runloop Trust Center.

What benchmarks are included on the Basic plan?

The Basic plan includes Runloop's public benchmark suite: SWE-Bench Verified and SWE-smith. These run in managed Runloop infrastructure with no setup or hosting required on your end. Custom benchmarks built from your own repositories -- private codebases, domain-specific test suites, regression harnesses -- require the Pro plan.

What types of AI use cases run on Runloop?

Runloop serves teams building coding agents across automated code review, test generation, long-context debugging, RL-based code synthesis, and benchmark evaluation (SWE-Bench, Multi-SWE, custom suites). Customers span AI-first startups building developer tools, enterprise teams deploying autonomous agents to production, and research labs running large-scale agent evaluations. The common thread is agents that need persistent execution environments with fast snapshot cycles -- not just a function runner.

Is Runloop suitable for both individual developers and enterprise teams?

Yes. The Basic plan supports individual developers and small teams exploring agent infrastructure with no upfront cost. Pro is designed for teams running agents in production with custom benchmarks and higher concurrency. Enterprise covers large organizations with compliance requirements, VPC deployment needs, and high-volume workloads that benefit from custom pricing and dedicated support.

How do I integrate Runloop with an existing agent pipeline?

Integration is straightforward through Runloop's Python and TypeScript SDKs and REST API. Devboxes expose standard shell and filesystem interfaces -- you bring your agent framework, Runloop provides the execution and evaluation substrate beneath it. Most teams go from API key to running their first devbox within a few hours. Full documentation at docs.runloop.ai.

Work at ease
Pay as you use

Try Our Pro Plan for Free!

Choose your plan

Usage

Everything You Need to Know

Get Started With Runloop

Get Started With Runloop

Work at easePay as you use

Try Our Pro Plan for Free!

Choose your plan

Usage

Everything You Need to Know

Get Started With Runloop

Get Started With Runloop

Work at ease
Pay as you use