See agent cost before it turns into invisible overhead.

ProPR is free, and it never marks up a token. You run planning, task history, logs, pull requests, and follow-ups in your own deployment, then pay providers directly through subscriptions or metered API billing. Provider charges go directly to your account.

Practical boundary

ProPR itself is free. The underlying provider usage — whether through subscriptions you already pay for or metered API billing — is the only cost beyond your own infrastructure.

The headline subscription covers only part of AI development cost. Other costs include long contexts, repeated retries, premium models, background agents, manual PR preparation, branch cleanup, review follow-ups, and re-establishing context after a private agent session is gone.

ProPR makes those choices explicit. A task has a plan, selected agent, isolated execution record, pull request, logs, usage, and follow-up history. When Agent Tank is enabled, ProPR ties subscription and rate-limit information back to the change that consumed it.

The bill has three parts.

ProPR (free)

The self-hosted product layer: plans, tasks, GitHub integration, PR follow-up, logs, settings, CLI, and task history. No license fee, no seat charge.

Your infrastructure

The Docker-capable host, storage, database state, repositories, worker capacity, and operational monitoring you choose to run.

Agent usage

The model or agent spend behind each task. Use your existing AI provider subscriptions (no additional cost) or direct API billing for metered accounting.

Spend the stronger model where it pays off.

A planning pass, careful implementation, inexpensive review, and PR follow-up can use different providers. ProPR keeps the branch and history connected while the agent changes.

ProPR settings showing a different model configured for each phase: GPT-5.4 Mini for plan context analysis, GPT-5.5 for plan generation, GPT-5.4 Mini for summarization, and Claude Opus 4.8 for pull-request review, with a system-wide reasoning level. — Real settings from a running stack: cheap models handle context analysis and summaries, a stronger model writes the plan, and review runs on a different provider entirely.

Use subscriptions you already have

Claude Code, Codex, Antigravity CLI, and other supported agents can use provider subscriptions directly. ProPR adds no token markup; provider plan limits still apply.

No ProPR token markupQuota-awareGood for routine tasks

Use direct APIs when metering matters

For usage that needs provider-level billing, model-specific control, or clearer per-task accounting, bring your own API keys instead.

Clearer spendProvider controlsTask-level logs

Choose the model by phase

Use the model that fits the phase, then keep follow-up commits, retries, and review history on the same PR.

PlanningImplementationReviewFollow-up

Show cost at the task level.

ProPR records every model call — model, token usage, cost, and duration — and ties it back to the task and repository that spent it, so agent spend is never invisible.

A focused ProPR usage log crop showing model, token and cost usage, and duration for recent model calls.

Compare against the full delivery cost.

A low monthly seat price can still be expensive if the team spends time turning private agent output into branches, PR summaries, follow-up prompts, and recovery steps.

Ask who pays for retries and long-running tasks.
Ask whether review follow-ups stay on the same branch.
Ask whether usage can be tied back to the change that consumed it.

Compare ProPR in market context.

The cost comparison places ProPR beside hosted agents, AI IDEs, app builders, open-source orchestration, direct agents, and PR review tools.