What actually differs between Quick, Standard and Deep?
The official page only says routine repos suit a quick pass and critical ones a deep review; it does not describe what each level actually does, or any difference in time and cost.
The third-party write-ups I found do not fill in those details either, so any specific difference figures should be doubted for now.
Can I use it on a Pro or personal account?
The official page says it is available to Claude Team and Enterprise organizations, enabled by admins. I found no description for individual plans.
Whether it is available to you depends on your plan and what the admin console shows.
How do I claim the $10,000 credits?
The official page says eligible Enterprise organizations can claim at least $10,000 by January 15, 2027, expiring 30 days after claiming, with terms applying.
For eligibility conditions and the claiming process, see the terms and support pages linked from the official page.
Does faster review mean better quality?
Not necessarily. Anthropic says it is faster and finds more of the bugs that matter, but gives no detection rate or precision, so the quality improvement is an official description rather than verifiable data.
The most reliable test is how many findings your team actually acts on, compared with before.
On October 15, 2026, Anthropic announced that Code Review is out of research preview and available to Claude Team and Enterprise organizations; the Claude Code v2.1.304 changelog of October 19 adds that Owners can choose Quick, Standard (the default) or Deep for the whole organization, and set a level for individual repositories. This article collects what is known and what Anthropic does not make clear.
The official page says most reviews now finish about 3x faster, from roughly 20 minutes down to 6 to 10 minutes (approximate figures, compared with the preview); reviews find more of the bugs that matter and state the problem, what triggers it and the fix in a few sentences. On levels, admins choose Quick, Standard (default) or Deep for each repository, with routine codebases getting a quick pass and critical ones a deep review, and the page stresses that every level delivers high-quality reviews. To enable it, admins turn on Code Review in Claude Code settings, install the GitHub App and select repositories; new PRs are then reviewed automatically with no setup for developers. For Enterprise, eligible organizations can claim at least $10,000 in Code Review credits by January 15, 2027, expiring 30 days after claiming, with terms applying.
The official page does not describe what each of the three levels does differently, gives no per-review price, no difference in time or cost between levels, and no concrete metrics such as detection rate or precision. The third-party write-ups I found do not fill these gaps either. So what can be said for certain is that there are three levels and they can be set per repo; what cannot be said is how many more problems Deep finds than Standard, or how much more it costs in money and time. Any article that states specific difference figures should be treated with suspicion until Anthropic publishes them.
With no official difference table, the sturdiest approach is to measure it yourself. Pick two repositories you know well that have known issues, keep one on Standard and set the other to Deep for now, run comparable PRs through both for a week or two, and record each review's duration, the number of findings, and how many your team actually acted on. Anthropic's suggested direction (Deep for critical repos, Quick for routine ones) can serve as a starting hypothesis, but do not treat it as a conclusion before you have data. On credits, the $10,000 must be claimed by the deadline and expires 30 days after claiming, so time the claim to match your expected usage.
If you are a Team or Enterprise admin, do two things first: check whether your organization is eligible for the credits and put January 15, 2027 on the calendar; then run the small two-repo comparison above before deciding which repositories deserve Deep. Until Anthropic publishes the cost of each level, treat setting everything to Deep as a budget risk to verify, not a default.