What changed with Cursor Bugbot billing on May 30, 2026?

Cursor announced that Bugbot is moving to usage-based billing for Teams and Individual plans, removing seat fees and adding configurable effort levels for deeper PR reviews or custom review logic. The change takes effect at the next renewal after June 8, 2026. Existing customers can opt in early. Practically: instead of paying per seat regardless of PR volume, you'll pay per review with effort level controlling depth (and cost). Heavy reviewers will pay more; light reviewers will pay less.

How does Cursor Bugbot compare to OpenAI Codex Cloud PR review?

Cursor Bugbot reviews PRs from inside Cursor and on GitHub via a Bugbot app, focused on code-quality and bug-spotting with configurable effort levels. OpenAI Codex Cloud runs code review as part of its container-isolated task model — submit a PR, Codex reviews in a sandboxed environment, returns comments and proposed fixes. Bugbot is the more turn-key SaaS review tool; Codex Cloud is the more powerful platform for review-plus-fix workflows but requires more setup.

How does Claude Code /code-review compare?

Claude Code's /code-review command (which replaced /simplify in May 2026 patches) runs Opus 4.8 against a target diff or branch, returning structured review comments tied to specific lines. It's local-first, integrated with whatever editor you have, and powered by the SWE-bench leader (88.6% Verified). It's not a managed SaaS — you bring your own Claude credentials and run it from terminal or via your editor's Claude Code integration. Best for teams already deep in the Claude Code workflow.

Which AI PR review tool should I use in June 2026?

For teams already in Cursor with bursty PR volume: Cursor Bugbot's new usage-based billing is the most cost-rational pick. For teams that want autonomous review-plus-fix in isolated containers: OpenAI Codex Cloud. For teams running Claude Code locally with strong opinions on permissions: Claude /code-review. Many teams use a mix — Bugbot or Codex for first-pass scanning, then a human plus Claude /code-review for the deeper look on critical changes.

Quick Answer

Cursor Bugbot Usage-Based vs Codex PR Review vs Claude Code Review (June 2026)

Published: June 1, 2026

Cursor Bugbot Usage-Based vs Codex PR Review vs Claude Code Review (June 2026)

On May 30, 2026, Cursor announced that Bugbot is moving to usage-based billing for Teams and Individual plans, removing seat fees and adding configurable effort levels. The change takes effect at the next renewal after June 8, 2026.

That’s a meaningful shift in how AI PR review is priced. It also opens up the question: how does Bugbot’s new model compare to OpenAI Codex Cloud’s review workflow and Claude Code’s /code-review command? Here’s the breakdown.

Last verified: June 1, 2026.

TL;DR

	Cursor Bugbot (usage-based)	OpenAI Codex Cloud review	Claude Code /code-review
Pricing model	Per-review, effort-based	Per-task in container	Per-token API + your sub
Where it runs	Cloud + GitHub app	OpenAI cloud (containers)	Local, your Claude API
Underlying model	Cursor-routed (often Claude / GPT)	OpenAI models (GPT-5.5, Codex variants)	Opus 4.8 (or Sonnet)
Review-only or review-plus-fix	Review-focused	Both (autonomous fix in branch)	Review-focused (local fix easy)
Effort level configurable	Yes (new in June 2026)	Per-task config	Per-prompt
Best for	Cursor-native teams	Container-isolated autonomous review	Audit-friendly local review

What Cursor Bugbot does (and why the billing change matters)

Bugbot is Cursor’s PR review product. It:

Sits as a GitHub app on your repos
Comments on PRs with line-tied review findings
Routes to Cursor’s underlying agent stack (Claude, GPT-5.5, Composer 2.5 — model choice depends on plan and task)
Integrates with Cursor’s IDE so you can jump from a Bugbot comment to a Cursor agent session

The old billing model (until June 8, 2026 renewals): per-seat. Every team member who could use Bugbot was a seat, regardless of how many PRs they reviewed.

The new billing model: usage-based. You pay per review. Effort level is configurable — a “quick scan” review costs less; a “deep review with custom logic” costs more.

Why this matters:

Teams with bursty PR volume (heavy weeks, light weeks) save money
Teams that mostly use Bugbot for big quarterly refactors don’t pay for unused months
Teams can dial up effort on critical PRs and dial down on routine ones

Effective at the next renewal after June 8, 2026. Existing customers can opt in early to start saving sooner.

What OpenAI Codex Cloud review does

Codex Cloud (May 2026 active) treats PR review as a task in an ephemeral container:

You hand Codex a PR or branch
Codex spins up a container with the repo, network/fs constrained
The agent runs review tools, reads the diff in context, optionally drafts fixes
Output: comments on the PR plus, optionally, a follow-up branch with proposed fixes

This is review-plus-fix, not just review. Codex can autonomously open a fix PR if you configure it that way. The container boundary means Codex can run arbitrary review tooling without you worrying about it touching your laptop.

The downside: Codex Cloud is OpenAI-cloud-only and US-cloud-dependent. For EU or regulated environments, this is often a blocker. And the per-task billing model can be expensive if you review thousands of PRs.

What Claude Code /code-review does

Claude Code’s /code-review slash command replaced /simplify in a May 2026 patch. It:

Runs in your local Claude Code session
Targets a diff, branch, or set of files
Powered by whatever Claude model you’ve configured (Opus 4.8 standard or fast mode)
Returns line-tied structured comments

Strengths:

Audit-friendly — runs against your local repo with your local permissions; no cloud container
High quality — Opus 4.8 is the SWE-bench Verified leader (88.6%)
Tight loop — if you see a comment, you can ask Claude Code to fix it in the same session
No new SaaS — uses your existing Claude credentials

Weaknesses:

Not a GitHub app — you’d need to script the GitHub integration yourself, or use a third-party glue
No multi-reviewer queue management — single-developer review tool, not a team SaaS

Cost math: a typical team

Mid-sized team, 20 engineers, ~200 PRs/month, average 250 lines per PR:

Cursor Bugbot (usage-based, post-June 8, 2026):

200 reviews/month at “standard” effort, ~$0.50 each = ~$100/month
Plus higher-effort reviews on 20 critical PRs at $2 each = $40/month
Total: ~$140/month

OpenAI Codex Cloud review:

200 review tasks/month at container cost ~$1.50–$3 each = $300–$600/month
Higher-quality output, especially with autonomous fix branches
Total: ~$400/month

Claude Code /code-review:

200 reviews/month, ~10K tokens in / ~3K tokens out at Opus 4.8 standard
Per-review cost ~$0.125 = ~$25/month
But each engineer needs a Claude Pro or Max subscription, so add ~$200–$2000/month base
Total: $225–$2025/month depending on plan mix

(Numbers approximate, will vary by repo size and effort level.)

Which to pick

Cursor Bugbot (post-June 8):

You’re already a Cursor team
PR volume is bursty
You want a GitHub-native managed SaaS
You like the per-review pricing transparency

OpenAI Codex Cloud:

You want review-plus-autonomous-fix
Container isolation matters for your security posture
You’re OK with US-cloud and OpenAI ecosystem lock-in
You have budget for higher-cost-per-review and want the highest autonomy

Claude Code /code-review:

Your engineers are already using Claude Code daily
You want local-first, audit-friendly review
You don’t need GitHub-app convenience (or you’ll script it)
You want the SWE-bench leader (Opus 4.8) on every review

Hybrid pattern

Many sophisticated teams in 2026 don’t pick one. A common pattern:

First-pass scan: Bugbot or Codex Cloud on every PR — catches obvious bugs, style issues
Deep review: Claude Code /code-review on critical PRs — runs locally, full Opus 4.8 quality
Human review: Senior eng signs off after AI passes

This minimizes cost (Bugbot is cheap on bulk), maximizes quality (Claude on what matters), and keeps a human in the loop where it’s needed.

Sources

Releasebot: Cursor release notes (Bugbot billing change, May 30, 2026) — usage-based detail
Cursor changelog — official Cursor 3.5 / 3.6 / Bugbot notes
Anthropic: Claude Code release notes — /code-review, May 2026 patches
Anthropic: Claude Opus 4.8 announcement

Bottom line

Cursor’s June 8 switch to usage-based Bugbot billing is the right move for bursty teams and a savings unlock for everyone who wasn’t reviewing PRs every day. But the broader question — Bugbot, Codex Cloud, or Claude /code-review? — depends on your existing stack and what tradeoff you want between automation, audit-friendliness, and cost.