Qodo

Qodo (formerly CodiumAI, founded 2022) is an AI code-review and code-quality platform: an agent that reviews pull requests on GitHub, GitLab, Bitbucket, Azure DevOps and Gerrit, enforces team-defined rules, and can generate tests for the code it flags. It ships as a GitHub/GitLab app for PR review, plus IDE extensions (VS Code, JetBrains) for local review before a PR is even opened, and a CLI ("Qodo Command") for scripting custom review agents into CI.

The VS Code extension (published under the verified 'Codium' / Qodo publisher) has 896,244 installs and a 4.69/5 average from 491 ratings as of today, per the Marketplace API. The JetBrains plugin has 644,239 downloads, per the JetBrains Marketplace API. Both are independently queryable and not vendor-controlled numbers. Qodo also lists Intel, HiBob, Nvidia, Macmillan, monday.com, Sansan, Walmart, Intuit and Box as customers on its pricing page — these are vendor-displayed logos, not corroborated by third-party case studies, so treat them as claims.

Qodo publishes its own 'AI Code Review Benchmark' claiming the strongest F1 score (precision/recall balance) among evaluated AI review tools, using injected bugs in real merged PRs from open-source repos. This is Qodo's own methodology and dataset — a self-published benchmark, not a third-party evaluation, and should be read as a vendor claim. Independent review-site data (G2, Gartner Peer Insights) could not be verified directly: G2's pages return a CAPTCHA/403 wall to both direct fetch and proxy access, and there is no Trustpilot listing for Qodo. Search-engine summaries cite G2 ratings in the 4.6–4.8/5 range and Gartner Peer Insights at 4.6/5 from 35 reviews, but these numbers come from secondary summaries, not a page we could load ourselves, so we're not treating them as established.

There is no permanent free tier. Qodo offers a 14-day trial with unlimited reviews and no credit card required. The paid plan, 'Pro Team,' is not per-seat: it's a shared credit pool (up to 30 users) priced at $0.012/credit, starting at $30/month for 2,500 credits (~18 reviews/month), scaling to $60/month for 5,000 credits and $240/month for 20,000 credits. Credits expire at the end of each monthly cycle and don't roll over. Once a team's base credits run out, reviews keep running at the same per-credit rate up to a self-set monthly overage cap, rather than stopping or charging a premium rate. Enterprise pricing is custom and adds SSO/SAML, audit logs, BYOK, single-tenant or on-prem deployment, and a dedicated CSM. Qualified open-source projects can apply for free access.

Qodo's early reputation was built on open-source tools, but that footprint has shrunk. PR-Agent — the original open-source PR-review agent and the project Qodo's commercial product grew out of — was transferred by Qodo in April 2026 to a new, community-governed GitHub organization ('The-PR-Agent'), reverting its license from AGPL-3.0 back to Apache 2.0 and adding an external maintainer. It is no longer a Qodo-controlled project (github.com/qodo-ai/pr-agent now redirects there), though it remains actively maintained (12,561 stars, last push within the last day). Separately, Qodo Cover — the AI-powered test-generation and coverage tool (AGPL-3.0, 5,612 stars) — was marked end-of-life by Qodo in June 2025 with a README notice telling users to fork it if they want to keep developing it; it has had no merged pull requests since June 2025. What remains actively maintained under the qodo-ai org today are smaller, mostly closed-license repos supporting the CLI and agent-workflow features (e.g., 'command', 'agents', 'open-aware'), not the review engine itself, which is closed-source SaaS.

Qodo was founded in 2022 (Itamar Friedman, CEO; Dedy Kredo, Chief Product Officer) as CodiumAI, rebranding to Qodo in 2024. It raised $11M in seed funding, then a $40M Series A in September 2024 led by Susa Ventures and Square Peg, bringing total disclosed funding to $50M. No funding announcement more recent than September 2024 was found as of this listing.

Qodo publishes a trust center (trust.qodo.ai) listing SOC 2 Type II certification and third-party penetration testing, along with a public Data Processing Addendum. The vendor states it does not train models on customer code. Enterprise plans add single-tenant or on-prem/air-gapped deployment and bring-your-own-key (BYOK) LLM options for teams that don't want code leaving their infrastructure at all.