Start free. Upgrade when you're ready.
Five review attempts on any public GitHub PR, no card required. Paid plans add private repos, per-review model selection, and unlimited standard-review volume — for one developer or a full team.
Cancel any time from your billing page. Free tier stays free.
Free
Try Veridu end-to-end on any public GitHub PR.
- 5 review attempts (lifetime, not per month)
- Public GitHub repos only
- Default reviewer panel (Codex + Claude Sonnet 5)
- Web, CLI, and hosted MCP surfaces
- Change-impact evidence on every review
- Reconciled findings across two vendors
Solo
For one developer working across private repos.
- Unlimited standard reviews (fair use)
- Public + private GitHub repos
- Full reviewer catalog with per-review model selection
- Everything in Free
- Priority email support
Team
Auto-review every PR across your organization.
- Unlimited standard reviews per seat (pooled, fair use)
- GitHub App auto-review on every PR
- Org-wide default panel + per-repo overrides
- Admin roles, member management, seat billing
- Slack / Linear / Jira destinations
- Everything in Solo
Enterprise
contact usFor teams that need SSO, self-hosting, or custom terms.
- Custom review volume + custom SLA
- Self-hosted option (your AWS / GCP)
- SSO (Okta, Google Workspace, Entra)
- Custom retention + audit-log exports
- Priority incident response
- Everything in Team
A standard review is the default flow: two AI reviewers, change-impact evidence up to a 50KB context budget, reconciled findings. Most PRs fit here comfortably.
Fair use means we don't cap paid standard reviews up front, and Dependabot / Renovate / generated-branch PRs don't count. We only step in for clear outlier usage — think hundreds of reviews per user per month. If that happens, we email first and work with you on plan fit or reviewer tuning. No silent throttling.
A future deep review mode - larger context, dependency tracing, tests + security passes - will be a separate metered add-on. Not shipping in the launch tiers.
No. We pay the vendor costs regardless of which models you pick from the catalog. One credential from Veridu covers the reviewer panel, the reconciliation engine, and the change-impact evidence pack.
One invocation of the reviewer pipeline against one pull request URL. Re-running against a force-pushed commit consumes another attempt. Invalid inputs (bad URL, private repo, quota exhausted) don't consume an attempt.
Three: install our GitHub App (recommended for teams — enables auto-review on every PR), add a fine-grained personal access token per repo (individual devs), or use the CLI in --local mode where nothing hits our servers. Pick one per repo; the /repos page shows what's linked.
No — see the data handling page. Prompts and diffs are sent to the reviewer models you selected; under their current published policies, neither OpenAI nor Anthropic uses API traffic for training. Veridu itself never trains on your code.
We keep review artifacts (diff, prompts, findings, verdicts) so you can revisit or export them later. Delete any review from the review list; delete your whole account from your billing page. Enterprise tier adds custom retention windows + audit-log export.
Monthly, via Stripe. Cancel or downgrade any time — the change applies at the end of your current billing period, and we don't prorate what you've already paid. Team plans adjust per seat added or removed. Enterprise terms are per-contract.
Email hello@veridu.io. We'd rather give a beta user more attempts than have you form an opinion off five reviews.