Introducing Augment Code Review, powered by GPT 5.2. It's the #1-ranked AI code reviewer across precision, recall, and overall quality. Free for the first week for every paying customer, and free for open source projects.

We benchmarked 7 leading AI code review tools across 50 real PRs using the only public dataset. Augment delivered the highest accuracy. It was also the only code review tool to achieve both high precision and high recall. This combination delivers high value for professional software teams: high signal, low noise.

Augment Code Review is powered by GPT-5.2, the latest model from @OpenAI. We chose GPT-5.2 because asynchronous code review demands different strengths than interactive coding: deeper reasoning, broader context retrieval, and the ability to spot subtle cross-system issues. GPT-5.2 excels at all three. It makes thorough tool calls, pulls in comprehensive context, and prioritizes correctness over speed—ideal for async review, where accuracy matters far more than latency. Read the blog:

Early users are already finding value. @mongoDB uses Augment Code Review to deliver cleaner, better-prepared code.

With Augment Code Review, @tekioncorp reduced their engineering team's average time to merge from 3 days to 1 day.

Having meaningful feedback from Augment Code Review reduced the @tilt team's review time, allowing engineers to focus on the most important things during the final human review.

Open source projects, like @dragonflydbio, with early access to Augment Code Review are already seeing meaningful impact. "The Augment Code Review agent has streamlined and boosted our development workflow. The ability to quickly identify and resolve issues directly within our existing processes has made our code reviews much more efficient and effective. It was surprising to see it work so well in a challenging C++ codebase like the Dragonfly project."

We're excited for you to try it! Open-source project? Get Augment Code Review free: Get your next PR reviewed:

@augmentcode dropped a 💣

🔥

This has caught so many bugs for me incredible addition to my workflow

Amazing news! Can’t wait to hear what it thinks of my code 😅

When will GPT-5.2 be available to select in the model selector for augmentation? I would like to use it.

Brilliant work 🫡 Just needs multi-org support for GitHub.

Are you guys banning accounts that use a VPN without warning them not to use a VPN? Seems there's an epidemic of Auggie accounts being unable to login ever again once a VPN is used.

Can't wait to see it available as one of the models for Agentic Coding.

And where is GPT 5.2 in extension? 🙄

Still running evals on 5.2 for the model picker - we'll keep everyone updated as we continue to test performance for general coding.

gpt5.2 gets confused by garbled code and indentation when writing Python, and gets lost in place.

Yes! Super excited to get this out to other people. It's a big value add.