Diffblue · Tools

Diffblue

Visit website

At a glance

Price
Pricing on request
Vendor
Diffblue

Specifications & properties

Key decision factors

Pricing model
Enterprise 1
Tool form
  • Code review
  • CLI agent
1
Autonomy level
Autonomous agent 1

Model

Available models
Orchestrates GitHub Copilot and Claude Code 1

Capabilities

Repo-wide context
Yes 1

Compliance

Self-hosting / on-prem
Yes 1
Report data / suggest a correction

Diffblue's Testing Agent autonomously generates, verifies, and commits unit tests for Java and Python across entire codebases.

Profile

Diffblue Cover is an AI-powered testing tool that automatically writes unit tests for Java and Kotlin code, using reinforcement learning to generate, run, and refine test candidates until they compile, pass, and add real coverage. It is built by Diffblue, a company that spun out of the University of Oxford's AI research group in 2016 and has focused exclusively on automated software testing ever since.

Who builds it

Diffblue was founded in 2016 by Daniel Kroening and Peter Schrammel, drawing on roughly a decade of Oxford research into program analysis; Mathew Lodge has been CEO since July 2019. The company raised a Series A round (reported as $22M / £17.3M) led by Goldman Sachs Principal Strategic Investments in 2017, with later backing from IP Group and AlbionVC. Diffblue counts large regulated enterprises among its customers, including Goldman Sachs, S&P Global, Citi, JP Morgan, and AWS.

How it works

Rather than suggesting code for a developer to accept or reject, Diffblue Cover works autonomously: it analyzes a Java or Kotlin codebase to map classes and methods, generates a test candidate for each testable method, runs the candidates, and uses reinforcement learning to iteratively adjust them until it converges on a suite that maximizes coverage. The whole process — plugin, CLI, and pipeline modes alike — runs inside the customer's own environment; Diffblue states that no code leaves the organization during test generation.

Core features

  • Autonomous unit-test writing for Java and Kotlin, including mocks and assertions, with no prompting required.
  • IntelliJ IDEA plugin for one-click test generation on classes and methods as you code.
  • Command-line interface that configures itself from an existing Maven or Gradle build.
  • CI/CD pipeline integration (GitHub, GitLab, Jenkins, Azure, AWS) to keep regression suites current automatically.
  • Reporting and, on higher tiers, additional Optimize and Refactor tooling for existing test suites.

Pricing

Diffblue Cover is sold in four editions:

  • Community — free, single user, up to 25 methods tested per month, IntelliJ plugin, community-forum support.
  • Developer — around $30/user/month (or roughly $330/year), single user, up to 100 methods per month.
  • Teams — roughly $30,000/year, 10+ user licenses, unlimited test generation for codebases up to 250,000 lines of code, plus CLI, pipeline, and reporting tools; a free trial is available.
  • Enterprise — custom pricing for 20+ licenses and codebases beyond 250,000 lines, adding Optimize/Refactor tooling and a dedicated support team.

Third-party listings (Capterra) are the source for the Developer and Teams list prices; Diffblue's own pricing page did not expose these figures in a form we could independently confirm, so treat exact numbers as indicative and check with Diffblue for a current quote.

Who it's for

Diffblue Cover targets Java-heavy engineering organizations — particularly regulated industries like banking and financial services, where its existing customer base sits — that need to raise or maintain unit-test coverage on large, legacy-heavy codebases without pulling developers off feature work. Individual developers can try the Community and Developer editions for free or at low cost; Teams and Enterprise are built for organizations running Cover across many repositories in CI/CD, where automatically-maintained regression suites reduce the manual test-writing burden at scale.

Frequently asked questions

Is Diffblue Cover free to use?

Yes, Diffblue Cover has a genuine free tier: the Community edition is free for a single user and lets you generate tests for up to 25 methods per month using the IntelliJ plugin. A paid Developer edition raises that cap to 100 methods per month for one user, and Diffblue also offers a free trial of the higher Teams edition for organizations that want to evaluate it on a larger codebase.

What does Diffblue Cover cost beyond the free tier?

Beyond the free Community edition, Diffblue Cover is sold through a Developer edition priced at roughly $30 per user per month (about $330 per year) for up to 100 methods monthly, a Teams edition listed at around $30,000 per year for 10+ users and unlimited test generation up to 250,000 lines of code, and a custom-quoted Enterprise edition for larger codebases and 20+ licenses. Diffblue's own pricing page did not expose exact figures for us to confirm directly, so treat these third-party-listed numbers as indicative and confirm current pricing with Diffblue's sales team.

What IDEs and platforms does Diffblue Cover support?

Diffblue Cover ships as a plugin for IntelliJ IDEA, generating tests for a class or method with one click as you write code. It's also available as a command-line tool that configures itself from an existing Maven or Gradle build, plus CI/CD pipeline integrations (GitHub, GitLab, Jenkins, Azure, AWS) so regression tests can be regenerated automatically as code changes.

Is Diffblue Cover open source, and does my code ever leave my environment?

Diffblue Cover is proprietary software, not open source — even the free Community edition is a closed-source product that is simply free to use. Diffblue states, however, that all test generation runs entirely inside the customer's own environment and that no code is sent externally during that process, which matters for teams in regulated industries.

What programming languages does Diffblue Cover support?

Diffblue Cover is built specifically for Java and Kotlin projects — it is not a general-purpose, multi-language tool. That focus is what lets its reinforcement-learning engine map a codebase's classes and methods precisely enough to generate compiling, passing unit tests without manual prompting.

Who builds Diffblue Cover, and which companies use it?

Diffblue Cover is built by Diffblue, a company spun out of the University of Oxford's AI research group in 2016 by Daniel Kroening and Peter Schrammel, with Mathew Lodge as CEO since 2019. The company raised a $22M (£17.3M) Series A in 2017 led by Goldman Sachs Principal Strategic Investments, and its publicly named customers include Goldman Sachs, S&P Global, Citi, JP Morgan, and AWS — reflecting its focus on large, regulated Java shops.