mlr3proba
mlr3proba is shedding weight as its survival work moves into sibling packages
A side-by-side editorial comparison of Countly and vetiver — release velocity, themes, recent moves, and the top alternatives to consider.
Countly's LTS line is spending its releases on hardening the surfaces customers extend.
The 25.03 LTS and 24.05 branches are moving together, and the recent content is dominated by security and scoping work rather than features. The latest LTS rebuilds the api and frontend Docker images as multi-stage builds on Debian 13 with Node 24 so compilers and build tooling no longer ship, overrides fourteen vulnerable transitive dependencies, and replaces the unmaintained v8-sandbox behind custom code with isolated-vm. Earlier releases scoped internal event hooks to their own apps, fixed the consents table returning fields beyond the consent columns, and stopped dashboard widgets being copied by users without access to the referenced apps.
Posit's MLOps package went quiet for two years, then came back to keep up with recipes.
vetiver versions, deploys and monitors models: it pins a model, generates a plumber API around it, and writes the Dockerfile to run it. The visible release stream is bug fixes to plumber file generation, one prototype endpoint, and then a two-year gap between 0.2.5 in November 2023 and 0.2.6 in October 2025. The two releases since that gap are compatibility work — recipes' new input data prototype, support for probably, and all versions of xgboost.
The 25.03 LTS and 24.05 branches are moving together, and the recent content is dominated by security and scoping work rather than features. The latest LTS rebuilds the api and frontend Docker images as multi-stage builds on Debian 13 with Node 24 so compilers and build tooling no longer ship, overrides fourteen vulnerable transitive dependencies, and replaces the unmaintained v8-sandbox behind custom code with isolated-vm. Earlier releases scoped internal event hooks to their own apps, fixed the consents table returning fields beyond the consent columns, and stopped dashboard widgets being copied by users without access to the referenced apps.
The pattern across these releases is closing the gaps where a customer-supplied artefact — custom hook code, a copied widget, a projection on a request — could reach further than intended. That work is now touching the runtime itself, and the isolated-vm swap is a breaking change: custom code relying on setTimeout, setInterval or async completion fails with a logged error instead of running. The journey engine is the only place shipping genuinely new capability, and it is enterprise-only.
Expect follow-up releases to soften the custom-code migration, since the isolated-vm switch silently breaks any hook that awaited a timer, and further ab-testing work now that pystan has been replaced with compiled Stan executables.
vetiver versions, deploys and monitors models: it pins a model, generates a plumber API around it, and writes the Dockerfile to run it. The visible release stream is bug fixes to plumber file generation, one prototype endpoint, and then a two-year gap between 0.2.5 in November 2023 and 0.2.6 in October 2025. The two releases since that gap are compatibility work — recipes' new input data prototype, support for probably, and all versions of xgboost.
The feature era ended before this window opened. Deploying to SageMaker, generating Docker files, storing renv lockfiles in model metadata and supporting keras, luz and recipes all landed in 0.2.1 and 0.2.2; nothing since has extended what vetiver does. What it does now is track the rest of tidymodels — when recipes gains a prototype API or probably becomes something a workflow can contain, vetiver adds a line. That is a package holding its position rather than advancing it.
The entries do not support a confident prediction of new capability. On this pattern the next release tracks another tidymodels change, most likely the postprocessing stage that workflows added in 1.3.0.
Other Analytics products tracked by Sparkpulse, ranked by recent ship velocity. Each card links to a full editorial trajectory and lets you pivot into a head-to-head comparison with either Countly or vetiver.
mlr3proba is shedding weight as its survival work moves into sibling packages
mlr3viz keeps the ecosystem's plots working while the plots themselves move out
mlr3tuning is rebuilding its async machinery under a stable public surface
timetk swallowed anomalize whole, then went quiet for two years
modelbased is turning marginal effects into a full contrast grammar
easystats' parameters package absorbs one more model class every few weeks
See all Countly alternatives → · See all vetiver alternatives →
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Countly is currently shipping more aggressively (velocity 5.0 vs 0.0), with 0 editorial sparks in the last 30 days against 0. See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Countly is currently shipping more aggressively (velocity 5.0 vs 0.0), with 0 editorial sparks in the last 30 days against 0. For your specific use case, the alternatives sections above list other Analytics products to evaluate alongside.
Top Countly alternatives in Analytics are ranked by recent ship velocity. Browse the "Countly alternatives" section above for the current picks, or visit /alternatives/countly for the full list with editorial commentary on each.
Top vetiver alternatives in Analytics are ranked by recent ship velocity. Browse the "vetiver alternatives" section above for the current picks, or visit /alternatives/vetiver-r for the full list with editorial commentary on each.