The pull request outlived its obituaries and got a bigger job
More changes are flowing through pull requests than ever thanks to AI-generated code, but a human still needs to decide what ships.
CodeRabbit Blog
AI-powered code reviews and software development insights from CodeRabbit
Live Last read · last published · next check
Latest posts
Introducing CodeRabbit Security
CodeRabbit Security traces suspected vulnerabilities across the codebase, evaluates reachability and the conditions required for exploitation against code evidence, and proposes reviewable fixes in the pull request workflow.
Code is abundant. Judgment is scarce. We raised $143 million to help it scale.
CodeRabbit raised a $143 million Series C at a $1.5 billion valuation and is introducing Agentic Change Management, the control layer for software changes created by humans and agents.
Teaching NVIDIA Nemotron 3.5 Lightning to route code reviews
CodeRabbit worked with NVIDIA and Baseten to post-train NVIDIA Nemotron 3.5 Lightning for one of CodeRabbit's highest-volume routing tasks, improving route agreement while reducing estimated inference cost.
The three hidden attention taxes derailing the agentic SDLC
Reviewing every agent-generated change is becoming impossible. The next bottleneck is deciding what deserves attention, rebuilding context, and keeping code safe after merge.
Better models don't solve a judgment bottleneck
Code is now plentiful, but judgment about what deserves to merge is not. Better models sharpen what agents write, but they don't make that call.
Opus 5 for code review: Cleaner actionable comments, noisier overall
Opus 5 produced a cleaner actionable-comment stream, but caught fewer known issues and generated roughly four times the baseline's nitpicks. Here is where the model may fit — and where it does not.
Code is no longer the bottleneck. Understanding is.
Coding agents can expand change volume faster than teams can understand it. Review needs a shared path from intent to system behavior to code so people can keep shaping the system.
Close the loop after every merge: the agent that reviewed your PR can now follow through
Post-Merge Actions use pull request context to handle changelogs, documentation, tickets, and other work that should happen after merge.
GPT-5.6 Sol and Terra: Where they fit for coding agents and code review
OpenAI’s GPT-5.6 family includes capability tiers: Sol as the flagship model, Terra as the lower-cost option, and Luna as the fastest, lowest-cost tier.
The case for monorepos, and what they cost you
Monorepos keep shared code in sync but make every bad PR everyone's problem. Here's when to commit to one, when polyrepo wins, and how to review at scale.
Claude Sonnet 5 review: Should you switch?
A hands-on review of Claude Sonnet 5 after a week of real coding and code-review work: how it compares to Sonnet 4.6, what it costs, and who should upgrade.
Loop engineering: designing loops you can actually walk away from
Loop engineering allows you to step away completely by designing a system that operates autonomously, removing the human from the loop entirely.
How a hackathon project turned into my work at CodeRabbit
How Ayush Sridhar CalHacks Hackathon project turned into an SWE internship with CodeRabbit
Automatic repository linking: cross-repo context without manual setup
CodeRabbit can detect related repositories across your organization and use them as review context, so cross-repo impact is easier to catch before you merge.
Humans don’t have an API
As generative AI becomes embedded in daily work, the line between how we communicate with software and how we communicate with one another can feel less distinct than it once did.
Fable 5 model review: early signals from code review and coding tasks
Fable 5 is worth testing for autonomous coding work, especially when the prompt is incomplete and the agent has to discover the environment before it can build. For production code review, the current baseline and Opus 4.8 still look safer.
CodeRabbit now supports NVIDIA Nemotron 3 Ultra
NVIDIA Nemotron 3 Ultra delivers accurate and fast throughput in CodeRabbit's self-hosted AI code reviews.
Nemotron 3 Ultra makes the case for fast, open coding models
Nemotron 3 Ultra brings fast open-weight reasoning to dev workflows. CodeRabbit benchmarks show near-baseline review performance with retries and validation.
You’re addicted to AI code generation. Now what?
Developers distrust AI coding tools just enough to double-check the output, yet rely on them too much to turn them off. Here's what that dependency is actually costing engineering teams, and how to build review systems that keep up with it.
Opus 4.8 benchmark results for AI code review and code generation
Opus 4.8 is the best model we have used for long-horizon agentic coding and code generation, and it holds its own on code review out of the box.
CodeRabbit is now in the Claude Marketplace
Anthropic customers can now apply their existing Anthropic spend commitment toward CodeRabbit.
AI can migrate your entire codebase. Reviewing it is another story
AI has dramatically lowered that barrier, and the bottleneck is shifting toward something else entirely: reviewing the large, complex diffs that migrations inevitably produce.
The diff says 1,400 lines. The change is six
CodeRabbit Review now has a semantic diff view that groups moved blocks and surfaces token-level edits, so the actual change isn't buried inside 1,400 lines of noise.
Explainable reviews: CodeRabbit Review and the context engine that make it possible
Learn how CodeRabbit delivers explainable AI code reviews through semantic diffs, context engineering, model orchestration, and verification systems built for enterprise software teams.
A major walkthrough upgrade: explainable PRs and smarter reviewer routing
CodeRabbit's PR walkthrough now organizes changes into logical layers and routes reviews to the right people automatically, so you spend less time reconstructing what happened and more time actually reviewing it.
Introducing Change Stack: The first AI-native code review interface
Change Stack turns any pull request into a guided walkthrough: logical change groups, inline diagrams, and layer-by-layer navigation. Built for big PRs.
Nobody is going to read the code
AI writes more code than humans can review. Logic errors are up 75%. Security issues nearly triple. Here's what replaces human review.
Introducing CodeRabbit Reverse Tunnel: AI code review for private-network enterprises
Connect CodeRabbit to private GitHub Enterprise Server, GitLab, and Gerrit instances with no inbound firewall rules, no IP allowlisting, and no PrivateLink. One outbound HTTPS connection. That's it.
How we built the CodeRabbit plugin for Codex
How CodeRabbit built their Codex plugin — lessons on skill design, auth handling, CLI vs app output, and keeping code review inside the developer flow.
What the Vercel breach means for enterprise code security
A stolen OAuth token brought down Vercel's internal systems. Learn the three security lessons every enterprise should take from this developer supply chain attack.
What changed in OpenAI GPT-5.5: better judgment, stronger coding, better signal
GPT-5.5 benchmark results from CodeRabbit show improved code review precision, higher signal, and better performance in real workflows.
Your AI agent has amnesia
Every coding agent session starts from zero. Learn why AI agent amnesia is quietly undermining team productivity, compounding context costs, and reversing 50 years of collaborative engineering progress — and what the next generation of agents needs to fix it.
What Claude Opus 4.7 means for AI code review
Claude Opus 4.7 outperformed in 100 evaluations across real open-source pull requests—finding more real bugs, delivering more actionable feedback, and reasoning across files better than anything we’ve tested.
Introducing the CodeRabbit plugin for Codex
Get AI-powered code reviews without leaving Codex. The CodeRabbit plugin runs reviews inside your session, catches bugs before PRs, and requires zero workflow changes to set up.
Our settings page was overwhelming our users. Here's what we did to fix it
our settings page became a wall of options that overwhelmed a lot of users. Here's how we solved it and what we learned along the way.
Gemini 3.1 Pro for code-related tasks: more focus, higher signal-to-noise
In practice, developers experience AI code review through the comments it leaves on pull requests: how often it finds real issues, how much noise it produces, and how actionable its feedback is. To an
Faster AI code reviews with NVIDIA Nemotron 3 Super
TL;DR: NVIDIA Nemotron 3 Super delivers high accuracy and faster throughput in CodeRabbit's self-hosted AI code reviews. We are happy to share that CodeRabbit is expanding its support for the NVIDIA N
Introducing one of the most requested CodeRabbit features: multi-repo analysis.
If you've ever merged a pull request that passed every check, looked clean in review, and then broke a downstream service ten minutes later…you already know the problem. When your architecture spans m
A semantic history: How the term 'vibe coding' went from a tweet to prod
A year is an eternity in AI time. In February 2025, Andrej Karpathy dropped a tweet-sized cultural marker into the software world: “vibe coding.” The phrase stuck because it captured a visceral shift
CodeRabbit tops the first independent AI code review benchmark
AI code review benchmarks have mostly been published by other code review vendors (whose tools always seem to come out on top in their benchmarks). We've written before about why we don’t think vendor
AI is burning out the people who keep open source alive
Over the past few months, one refrain has been heard consistently in open source communities: “AI slop.” It shows up in LinkedIn discussions from CEOs who rarely complain publicly. It shows up in Re
Developers are dead? Long live developers.
Predictions about the end of programming are nothing new. Every few years, someone confidently announces that this time developers are truly finished. If you listened to these self-proclaimed Nostradamuses, devs were previously set to be replaced by ...
Issue Planner: collaborative planning for teams using coding agents
For decades, the software development lifecycle has followed a familiar timeline. You create an issue, assign the work, manually write the code, get several peers to review it, test it, and ship. Each step took a relatively predictable amount of time...
Misalignment: The hidden cost of AI coding agents isn't from AI at all
TL;DR: The real cost of AI agents isn’t tokens or tools; it’s misalignment that shows up as rework, slop, and slowed teams. The conversation everyone is having (and why it misses the point) Most conversations about AI coding agents sound like a fant...
We are committed to supporting open source: distributed $600,000 to open source maintainers in 2025
CodeRabbit recognizes the growing need to support open source software (OSS), especially as AI accelerates the development landscape. While AI makes writing code faster and increases the frequency of pull requests, the time and effort of maintainers ...
An (actually useful) framework for evaluating AI code review tools
Benchmarks promise clarity. They’re supposed to reduce a complex system to a score, compare competitors side by side, and let the numbers speak for themselves. But, in practice, they rarely do. Benchmarks don’t measure “quality” in the abstract. They...
Why users shouldn’t choose their own LLM models: choice is not always good
Giving users a dropdown of LLMs to choose from often seems like the right product choice. After all, users might have a favorite model or they might want to try the latest release the moment it drops. One problem: unless they’re an ML engineer runnin...
CodeRabbit's AI code reviews now support NVIDIA Nemotron
TL;DR: Blend of frontier & open models is more cost efficient and reviews faster. NVIDIA Nemotron is supported for CodeRabbit self-hosted customers. We are delighted to share that CodeRabbit now supports the NVIDIA Nemotron family of open models amon...
Our new report: AI code creates 1.7x more problems
What we learned from analyzing hundreds of open-source pull requests. Over the past year, AI coding assistants have gone from emerging tools to everyday fixtures in the development workflow. At many organizations, a part of every code change is now m....