All posts
Product
·
5 min read
Introducing Sable 2.0: evals that run in CI
Prompt changes now go through review like code. Sable 2.0 scores every pull request against your golden datasets and blocks regressions before they merge.
Priya Natarajan
Head of Product
Prompts are code. Treat them that way.
The fastest-moving teams we work with change prompts, tools and models several times a day. Until now, the only way to know whether a change made things better was to ship it and watch the dashboards.
Sable 2.0 moves that feedback loop into your pull requests. Point the CLI at a dataset and an eval suite, and every PR gets a scored diff: what improved, what regressed, and the exact runs that changed.
What’s new
Sable CLI with GitHub, GitLab and Buildkite integrations
Configurable quality gates that fail the build on regressions
Side-by-side run diffs linked directly from the PR comment
40+ built-in scorers, plus custom evals in Python or TypeScript
Sable 2.0 is available today on every plan, including Hobby.