What is Qodo?
Qodo, the company formerly called CodiumAI, reviews pull requests. When someone opens a pull request on GitHub, GitLab, Bitbucket, Azure DevOps or Gerrit, a group of specialised agents reads the change against your repositories, past pull requests and written team rules, then posts findings and suggested fixes. Version 3.0, released on 1 October 2026, added a shared triage queue for open pull requests across repositories and an analytics dashboard; 3.2 followed on 2 November.
The product has narrowed this autumn. The IDE plugin for VS Code, JetBrains and Visual Studio was retired on 1 November 2026, and the old test-generation command was deprecated with no replacement. Local review now runs through the Agentic Toolbox, a CLI and MCP server that coding agents such as Claude Code, Codex and Kiro call before a pull request exists.
The open-source PR-Agent that Qodo started now lives in a separate community organisation on GitHub, with about 13,900 stars, and its README says it is not the Qodo free tier. Qodo raised a $70 million Series B in March 2026, bringing its total to $120 million.
How much does Qodo cost?
| Plan | Price | What you get |
|---|---|---|
| Trial | Free for 14 days | Unlimited reviews and credits, no card |
| Pro Team | From $30/month | 2,500 credits (about 18 reviews); 5,000 and 20,000 credit packs at $0.012 per credit; up to 30 users |
| Enterprise | Custom | 30+ users; SSO, audit logs, bring your own model keys, single-tenant, on-premises or air-gapped |
14-day free trial with unlimited reviews, no card needed; no permanent free tier. Pro Team (up to 30 users, monthly, no commitment) from $30/month for 2,500 credits (about 18 reviews), with packs of 5,000 or 20,000 credits; all credits $0.012 each, pooled across the team, with overage at the same rate up to a cap you set. Enterprise (30+ users) custom: SSO, audit logs, bring your own model keys, single-tenant, on-premises or air-gapped. Free for qualifying open-source projects.
Is Qodo worth it?
When a coding agent writes half your pull requests, the slow part of the week becomes reading them. A reviewer that knows your team's rules and catches the obvious problems first takes some of that weight off.
Qodo reads each pull request and leaves findings, using several specialised agents rather than one model pass. It draws on your other repositories, past pull requests and a set of written rules it can import from files already in your repos. Version 3.1, released on 15 October, switches off rules whose findings your team rarely fixes, which is the kind of noise control AI reviewers usually lack.
Where it pulls ahead of most rivals is where it can run. The docs list multi-tenant cloud, a single-tenant instance, on-premises, and fully air-gapped installs for GitHub Enterprise, GitLab Self-Managed, Bitbucket Data Center and Gerrit. CodeRabbit offers self-hosting only on Enterprise and does not support Gerrit; GitHub Copilot's reviewer works only on GitHub. For a bank or a defence supplier whose code may not leave the building, that table matters more than any benchmark.
On data, the pricing page says Qodo does not train models on your code and uses it only to produce your reviews. A 2023 security post on Qodo's blog lists SOC 2 Type II, and its retention policy is available to paying customers through the trust centre rather than published.
The catch: you buy credits, not seats, and a review costs whatever its size and complexity cost. Qodo's own guide says the $30 pack of 2,500 credits covers about 18 reviews a month, roughly $1.67 a review, against $1 for a Base review on Greptile. Unused credits expire each month. The product is also shrinking at the edges: the IDE plugin stopped working on 1 November 2026, with settings, chat history and custom agents not carried over, and the v1 test command, the feature CodiumAI was first known for, has no replacement. And by default Qodo silently skips pull requests opened by bot accounts, including ones named claude or codex, so an agent-heavy team has to switch that off before it reviews anything the agents write.
Our take: Qodo earns its place when the code cannot leave your network or lives on Gerrit, where few rivals will follow. For a cloud team on GitHub, CodeRabbit and Copilot do the same job with a price you can read off the page.
Where it earns its place
A platform lead at a European bank running GitLab Self-Managed
Add AI review without any code leaving our network
Enterprise covers an air-gapped install with bring-your-own model keys, SSO and audit logs; pricing is negotiated on an annual contract. The docs list GitLab Self-Managed as supporting every deployment model, including air-gapped.
A ten-person startup team on GitHub using Claude Code
Have every agent-written pull request checked before a person reads it
Pro Team starts at $30 a month for about 18 reviews, so a team opening 100 pull requests a month needs a larger pack or overage at $0.012 per credit. Set ignore_bot_pr to false first, or Qodo skips the agent's pull requests without comment.
An engineering manager at an automotive supplier on Gerrit
Enforce written coding standards across 40 repositories
Gerrit support is Enterprise-only. Rules are imported from files in the repos, applied at workspace, Git organisation or repository level, and rarely acted-on rules are switched off automatically unless locked.
Before you commit
- Setup
- Sign in with GitHub, Google or email and install the Qodo app on your repositories; admin access is needed, and reviews start on the next pull request. The Agentic Toolbox CLI needs Node.js 20 or later.
- Integrations
- GitHub, GitLab, Bitbucket, Azure DevOps and Gerrit (Enterprise); Jira, Linear and Azure DevOps tickets as review context; Agentic Toolbox as a CLI, an MCP server and plugins for Claude Code, Codex and Kiro.
- Data and privacy
- Qodo does not train models on your code, which is used only to generate reviews; SOC 2 Type II per Qodo's security blog; retention policy available to paid customers in the trust centre; single-tenant, on-premises and air-gapped deployment on Enterprise. Which AI tools train on your data?
- Cost at scale
- Pro Team caps at 30 users; at the published $0.012 per credit, the 20,000-credit pack (about 144 reviews) works out at $240 a month, and larger teams move to an annual Enterprise contract with custom pricing.
- Limits
- Credits per review vary with pull request size and are not published per review; credits expire monthly; bot-authored pull requests are skipped by default; comments are trimmed to each Git host's size limit.
What’s good
- On-premises and air-gapped installs, including GitHub Enterprise, GitLab Self-Managed and Gerrit
- Reviews on GitHub, GitLab, Bitbucket, Azure DevOps and Gerrit
- Switches off rules whose findings are rarely fixed
- Agentic Toolbox lets Claude Code, Codex and Kiro run the same review before a pull request
- No training on customer code; monthly Pro Team billing with no commitment
What’s not
- Credit pricing: about 18 reviews for $30, and unused credits expire monthly
- IDE plugin retired on 1 November 2026; settings and chat history not migrated
- Test generation command deprecated with no replacement
- Skips pull requests opened by bots and AI agents unless you change a setting
- No permanent free tier; SSO, audit logs and self-hosting need Enterprise
Best for
Regulated engineering teams that need an AI pull request reviewer running on their own infrastructure, including air-gapped and Gerrit set-ups.
Qodo alternatives
Qodo: common questions
- Is Qodo worth it?
- Qodo is a capable pull request reviewer that will run behind your own firewall, but it has just dropped its IDE plugin and test generation, and its credit pricing makes the monthly bill hard to predict. We rate it 3.8 out of 5.
- How much does Qodo cost?
- 14-day free trial with unlimited reviews, no card needed; no permanent free tier. Pro Team (up to 30 users, monthly, no commitment) from $30/month for 2,500 credits (about 18 reviews), with packs of 5,000 or 20,000 credits; all credits $0.012 each, pooled across the team, with overage at the same rate up to a cap you set. Enterprise (30+ users) custom: SSO, audit logs, bring your own model keys, single-tenant, on-premises or air-gapped. Free for qualifying open-source projects.
- What are the downsides of Qodo?
- Credit pricing: about 18 reviews for $30, and unused credits expire monthly; iDE plugin retired on 1 November 2026; settings and chat history not migrated; test generation command deprecated with no replacement; skips pull requests opened by bots and AI agents unless you change a setting; no permanent free tier; SSO, audit logs and self-hosting need Enterprise.
- Who is Qodo best for?
- Regulated engineering teams that need an AI pull request reviewer running on their own infrastructure, including air-gapped and Gerrit set-ups.
- How much does Qodo cost for a team or in production?
- Pro Team caps at 30 users; at the published $0.012 per credit, the 20,000-credit pack (about 144 reviews) works out at $240 a month, and larger teams move to an annual Enterprise contract with custom pricing.
- Is Qodo safe for company data?
- Qodo does not train models on your code, which is used only to generate reviews; SOC 2 Type II per Qodo's security blog; retention policy available to paid customers in the trust centre; single-tenant, on-premises and air-gapped deployment on Enterprise.
- What are the best Qodo alternatives?
- The closest alternatives we have reviewed are CodeRabbit, Greptile, GitHub Copilot, Cursor and Augment Code.
Answers checked Nov 2, 2026.