In this piece
Greptile Base vs Plus vs Apex: A PR Review Budget
Evidence: Greptile’s Plus and Apex launch post, Pro pricing, tier configuration and billing documentation, checked on 11 November 2026. The pilot below is a proposal; we have not run it or measured these tools ourselves.
Which Greptile tier should a team use?
Start with the consequences of a missed defect. A small authorization change can deserve more scrutiny than a large mechanical refactor. Our proposed rule is Base for bounded changes, Plus when correctness crosses module boundaries, and Apex when a missed issue could affect access, money, migrations or recovery. Validate this rule on your own pull requests before making it automatic.
Greptile’s 25 September 2026 launch announcement introduced Plus and Apex alongside Base. Its example demonstrates findings at different depths on one Celestia change. That is vendor evidence of a useful capability, not an independent estimate of your team’s defect recall.
What do Base, Plus and Apex cost?
Greptile Pro currently lists USD 1 per prepaid credit and a shared organization balance with no seat fee. Starter offers 50 free credits per month for one active developer. Each completed review and rerun draws its tier’s credits. Current Greptile pricing
| Tier | Credits per review | Proposed pilot allocation |
|---|---|---|
| Base | 1 | Bounded changes with straightforward acceptance tests |
| Plus | 3 | Cross-module behavior or uncertain call sites |
| Apex | 10 | High-impact changes needing deeper independent scrutiny |
Example organization budget. Thirty Base, five Plus and two Apex reviews cost 65 credits, or USD 65 at the listed Pro rate. Running all 37 at Apex costs 370 credits, or USD 370. Adding TREX to five completed runs adds 15 credits, bringing the first scenario to USD 80 before taxes or discounts. Compare confirmed findings and review time for each scenario.
How do monorepo settings change the budget?
Auto charges the tier that actually runs. In a monorepo, a PR that touches directories with different settings uses the highest tier. The CLI uses Base without an effort flag and is excluded from Starter. TREX works with all review tiers and adds three credits when at least one flow reaches a pass or fail. Review-tier configuration
Put these details into the evaluation manifest. Record the tier shown on each completed review, changed directories and trigger. Otherwise a comparison labeled “Base versus Apex” may actually compare different routing configurations. Do not assume a UI setting also governs a CLI pilot.
Credits are prepaid and shared across the organization. Record completed reviews, reruns and eligible TREX runs against the common balance. Monitor per-developer usage for attribution and check the balance and automatic recharge settings before a pilot. New reviews pause when credits run out. Greptile billing documentation
How can you test whether Apex is worth it?
Use a permission-cleared sample covering ordinary changes, cross-module changes and high-impact changes. Include clean PRs and PRs with independently confirmed defects. Fix the commit, repository context and review instructions; run each tier on the same starting point. Keep the known-defect answers out of the reviewer prompt.
- Label findings blind to tier, then reproduce each reported defect.
- Count distinct actionable defects, missed known defects and false positives. Several comments about one issue are one finding.
- Record elapsed review time, charged credits and human triage minutes.
- Repeat a subset to expose run-to-run variation. Report sample size and exclusions.
- Compare additional confirmed findings with additional cost and triage time within each risk group.
Avoid calling a fixed historical sample “recall on all bugs.” You only know the defects independently established in that sample. Findings without a reproduction remain unconfirmed, even when their language is persuasive.
What should the team still review manually?
Keep business invariants, migration recovery and permission boundaries in human review and executable tests. A deeper reviewer can surface a dependency path; it cannot decide whether your refund policy or tenant contract is correct. A clean review should not silently become a production release approval.
Track review cost per accepted PR separately from the cost per confirmed additional defect. The first measures operating cost; the second helps decide where escalation earns its budget. If Apex produces more comments but no additional verified findings in a group, keep investigating before expanding its use.
Does a high file count automatically justify Apex?
No. Greptile recommends Apex for production-targeted and large PRs, but file count is only a proxy for risk. A generated rename and a three-line ownership check have different failure consequences. Use path and risk rules together, then revisit them when the pilot reveals which changes benefit.
When should you adopt the deeper tiers?
Adopt them where additional confirmed findings justify credits and review delay. Keep cheaper paths where the same acceptance evidence is sufficient. Bring your review sample, defect labels and escalation rules to discuss a review-process pilot if the unresolved work is evaluating your codebase rather than choosing a subscription label.
If another reviewer returns nothing, our two-file Alibaba Open Code Review experiment shows how to check which changed files it selected.
Related implementation guidance
Graphify Review 2026: Is a Codebase Knowledge Graph Worth It?. Canary AI QA: Test Defect Detection, Not Benchmark Scores.
Sources checked
Independence and trademarks: Wavect publishes this page and is itself a provider, so we have a commercial interest in it. We are not affiliated with, endorsed by or partnered with the other companies named here, and all third-party company names, brands and trademarks are the property of their respective owners. Statements about other providers are taken from publicly available sources, primarily their own published pages, as of the review date shown on this page, and may have changed since. Please verify them directly before you decide. This page was written to the best of our knowledge and with the intent to remain objective. If you believe anything here is inaccurate or unfair, write to us and we will correct it: [email protected]
