1.5.0 - 05 Mar 2024

What’s Changed

  • Update CHANGELOG by @github-actions in

  • build: Migrate away from Node v16 actions by @jesse-c in

  • build: Bump version and improve release doc by @jesse-c in

  • build: Upgrade stale packages (fastapi, starlette, tensorflow, torch) by @sakoush in

  • fix(ci): tests and security workflow fixes by @sakoush in

  • Re-generate License Info by @github-actions in

  • fix(ci): Missing quote in CI test for all_runtimes by @sakoush in

  • build(docker): Bump dependencies by @jesse-c in

  • docs: List supported Python versions by @jesse-c in

  • fix(ci): Have separate smaller tasks for release by @sakoush in


  • We remove support for python 3.8, check for more info. Docker images for mlserver are already using python 3.10.

Full Changelog:…1.5.0


1.4.0 - 28 Feb 2024

What’s Changed

  • Free up some space for GH actions by @adriangonz in

  • Introduce tracing with OpenTelemetry by @vtaskow in

  • Update release CI to use Poetry by @adriangonz in

  • Re-generate License Info by @github-actions in

  • Add support for white-box explainers to alibi-explain runtime by @ascillitoe in

  • Update CHANGELOG by @github-actions in

  • Fix error when copying to output path by @lc525 in

  • Fix typo by @strickvl in

  • feat(logging): Distinguish logs from different models by @vtaskow in

  • Make sure we use our Response class by @adriangonz in

  • Adding Quick-Start Guide to docs by @ramonpzg in

  • feat(logging): Provide JSON-formatted structured logging as option by @vtaskow in

  • Bump in conda version and mamba solver by @dtpryce in

  • feat(huggingface): Merge model settings by @jesse-c in

  • feat(huggingface): Load local artefacts in HuggingFace runtime by @vtaskow in

  • Document and test behaviour around NaN by @adriangonz in

  • Address flakiness on ‘mlserver build’ tests by @adriangonz in

  • Bump Poetry and lockfiles by @adriangonz in

  • Bump Miniforge3 to 23.3.1 by @adriangonz in

  • Re-generate License Info by @github-actions in

  • Improved huggingface batch logic by @ajsalow in

  • Add inference params support to MLFlow’s custom invocation endpoint (… by @M4nouel in

  • Increase build space for runtime builds by @adriangonz in

  • Fix minor typo in sklearn README by @krishanbhasin-gc in

  • Add catboost classifier support by @krishanbhasin-gc in

  • added model_kwargs to huggingface model by @nanbo-liu in

  • Re-generate License Info by @github-actions in

  • Local response cache implementation by @SachinVarghese in

  • fix link to custom runtimes by @kretes in

  • Improve typing on Environment class by @krishanbhasin-gc in

  • build(dependabot): Change reviewers by @jesse-c in

  • MLServer changes from internal fork - deps and CI updates by @sakoush in

New Contributors

  • @vtaskow made their first contribution in

  • @lc525 made their first contribution in

  • @strickvl made their first contribution in

  • @ramonpzg made their first contribution in

  • @jesse-c made their first contribution in

  • @ajsalow made their first contribution in

  • @M4nouel made their first contribution in

  • @nanbo-liu made their first contribution in

  • @kretes made their first contribution in

Full Changelog:…1.4.0


1.3.5 - 10 Jul 2023

What’s Changed

  • Rename HF codec to hf by @adriangonz in

  • Publish is_drift metric to Prom by @joshsgoldstein in

New Contributors

  • @joshsgoldstein made their first contribution in

Full Changelog:…1.3.5


1.3.4 - 21 Jun 2023

What’s Changed

  • Silent logging by @dtpryce in

  • Fix mlserver infer with BYTES by @RafalSkolasinski in

New Contributors

  • @dtpryce made their first contribution in

Full Changelog:…1.3.4


1.3.3 - 05 Jun 2023

What’s Changed

  • Add default LD_LIBRARY_PATH env var by @adriangonz in

  • Adding cassava tutorial (mlserver + seldon core) by @edshee in

  • Add docs around converting to / from JSON by @adriangonz in

  • Document SKLearn available outputs by @adriangonz in

  • Fix minor typo in alibi-explain tests by @ascillitoe in

  • Add support for .ubj models and improve XGBoost docs by @adriangonz in

  • Fix content type annotations for pandas codecs by @adriangonz in

  • Added option to configure the grpc histogram by @cristiancl25 in

  • Add OS classifiers to project’s metadata by @adriangonz in

  • Don’t use qsize for parallel worker queue by @adriangonz in

  • Fix small typo in Python API docs by @krishanbhasin-gc in

  • Fix star import in mlserver.codecs.* by @adriangonz in

New Contributors

  • @cristiancl25 made their first contribution in

  • @krishanbhasin-gc made their first contribution in

Full Changelog:…1.3.3


1.3.2 - 10 May 2023

What’s Changed

  • Use default initialiser if not using a custom env by @adriangonz in

  • Add support for online drift detectors by @ascillitoe in

  • added intera and inter op parallelism parameters to the hugggingface … by @saeid93 in

  • Fix settings reference in runtime docs by @adriangonz in

  • Bump Alibi libs requirements by @adriangonz in

  • Add default LD_LIBRARY_PATH env var by @adriangonz in

  • Ignore both .metrics and .envs folders by @adriangonz in

New Contributors

  • @ascillitoe made their first contribution in

Full Changelog:…1.3.2


1.3.1 - 27 Apr 2023

What’s Changed

  • Move OpenAPI schemas into Python package (#1095)


1.3.0 - 27 Apr 2023

WARNING :warning: : The 1.3.0 has been yanked from PyPi due to a packaging issue. This should have been now resolved in >= 1.3.1.

What’s Changed

Custom Model Environments

More often that not, your custom runtimes will depend on external 3rd party dependencies which are not included within the main MLServer package - or different versions of the same package (e.g. scikit-learn==1.1.0 vs scikit-learn==1.2.0). In these cases, to load your custom runtime, MLServer will need access to these dependencies.

In MLServer 1.3.0, it is now possible to load this custom set of dependencies by providing them, through an environment tarball, whose path can be specified within your model-settings.json file. This custom environment will get provisioned on the fly after loading a model - alongside the default environment and any other custom environments.

Under the hood, each of these environments will run their own separate pool of workers.


Custom Metrics

The MLServer framework now includes a simple interface that allows you to register and keep track of any custom metrics:

  • [mlserver.register()]( Register a new metric.

  • [mlserver.log()]( Log a new set of metric / value pairs.

Custom metrics will generally be registered in the [load()]( method and then used in the [predict()]( method of your custom runtime. These metrics can then be polled and queried via Prometheus.



MLServer 1.3.0 now includes an autogenerated Swagger UI which can be used to interact dynamically with the Open Inference Protocol.

The autogenerated Swagger UI can be accessed under the /v2/docs endpoint.

Alongside the general API documentation, MLServer also exposes now a set of API docs tailored to individual models, showing the specific endpoints available for each one.

The model-specific autogenerated Swagger UI can be accessed under the following endpoints:

  • /v2/models/{model_name}/docs

  • /v2/models/{model_name}/versions/{model_version}/docs

HuggingFace Improvements

MLServer now includes improved Codec support for all the main different types that can be returned by HugginFace models - ensuring that the values returned via the Open Inference Protocol are more semantic and meaningful.

Massive thanks to @pepesi for taking the lead on improving the HuggingFace runtime!

Support for Custom Model Repositories

Internally, MLServer leverages a Model Repository implementation which is used to discover and find different models (and their versions) available to load. The latest version of MLServer will now allow you to swap this for your own model repository implementation - letting you integrate against your own model repository workflows.

This is exposed via the model_repository_implementation flag of your settings.json configuration file.

Thanks to @jgallardorama (aka @jgallardorama-itx ) for his effort contributing this feature!

Batch and Worker Queue Metrics

MLServer 1.3.0 introduces a new set of metrics to increase visibility around two of its internal queues:

Many thanks to @alvarorsant for taking the time to implement this highly requested feature!

Image Size Optimisations

The latest version of MLServer includes a few optimisations around image size, which help reduce the size of the official set of images by more than ~60% - making them more convenient to use and integrate within your workloads. In the case of the full seldonio/mlserver:1.3.0 image (including all runtimes and dependencies), this means going from 10GB down to ~3GB.

Python API Documentation

Alongside its built-in inference runtimes, MLServer also exposes a Python framework that you can use to extend MLServer and write your own codecs and inference runtimes. The MLServer official docs now include a reference page documenting the main components of this framework in more detail.

New Contributors

  • @rio made their first contribution in

  • @pepesi made their first contribution in

  • @jgallardorama made their first contribution in

  • @alvarorsant made their first contribution in

  • @gawsoftpl made their first contribution in

  • @stephen37 made their first contribution in

  • @sauerburger made their first contribution in


1.2.4 - 10 Mar 2023

Full Changelog:…1.2.4


1.2.3 - 16 Jan 2023

Full Changelog:…1.2.3


1.2.2 - 16 Jan 2023

Full Changelog:…1.2.2


1.2.1 - 19 Dec 2022

Full Changelog:…1.2.1


1.2.0 - 25 Nov 2022

What’s Changed

Simplified Interface for Custom Runtimes

MLServer now exposes an alternative “simplified” interface which can be used to write custom runtimes. This interface can be enabled by decorating your predict() method with the mlserver.codecs.decode_args decorator, and it lets you specify in the method signature both how you want your request payload to be decoded and how to encode the response back.

Based on the information provided in the method signature, MLServer will automatically decode the request payload into the different inputs specified as keyword arguments. Under the hood, this is implemented through MLServer’s codecs and content types system.

from mlserver import MLModel
from mlserver.codecs import decode_args

class MyCustomRuntime(MLModel):

  async def load(self) -> bool:
    # TODO: Replace for custom logic to load a model artifact
    self._model = load_my_custom_model()
    self.ready = True
    return self.ready

  async def predict(self, questions: List[str], context: List[str]) -> np.ndarray:
    # TODO: Replace for custom logic to run inference
    return self._model.predict(questions, context)

Built-in Templates for Custom Runtimes

To make it easier to write your own custom runtimes, MLServer now ships with a mlserver init command that will generate a templated project. This project will include a skeleton with folders, unit tests, Dockerfiles, etc. for you to fill.


Dynamic Loading of Custom Runtimes

MLServer now lets you load custom runtimes dynamically into a running instance of MLServer. Once you have your custom runtime ready, all you need to do is to move it to your model folder, next to your model-settings.json configuration file.

For example, if we assume a flat model repository where each folder represents a model, you would end up with a folder structure like the one below:

├── models
│   └── sum-model
│       ├── model-settings.json
│       ├──

Batch Inference Client

This release of MLServer introduces a new mlserver infer command, which will let you run inference over a large batch of input data on the client side. Under the hood, this command will stream a large set of inference requests from specified input file, arrange them in microbatches, orchestrate the request / response lifecycle, and will finally write back the obtained responses into output file.

Parallel Inference Improvements

The 1.2.0 release of MLServer, includes a number of fixes around the parallel inference pool focused on improving the architecture to optimise memory usage and reduce latency. These changes include (but are not limited to):

  • The main MLServer process won’t load an extra replica of the model anymore. Instead, all computing will occur on the parallel inference pool.

  • The worker pool will now ensure that all requests are executed on each worker’s AsyncIO loop, thus optimising compute time vs IO time.

  • Several improvements around logging from the inference workers.

Dropped support for Python 3.7

MLServer has now dropped support for Python 3.7. Going forward, only 3.8, 3.9 and 3.10 will be supported (with 3.8 being used in our official set of images).

Move to UBI Base Images

The official set of MLServer images has now moved to use UBI 9 as a base image. This ensures support to run MLServer in OpenShift clusters, as well as a well-maintained baseline for our images.

Support for MLflow 2.0

In line with MLServer’s close relationship with the MLflow team, this release of MLServer introduces support for the recently released MLflow 2.0. This introduces changes to the drop-in MLflow “scoring protocol” support, in the MLflow runtime for MLServer, to ensure it’s aligned with MLflow 2.0.

MLServer is also shipped as a dependency of MLflow, therefore you can try it out today by installing MLflow as:

$ pip install mlflow[extras]

To learn more about how to use MLServer directly from the MLflow CLI, check out the MLflow docs.

New Contributors

  • @johnpaulett made their first contribution in

  • @saeid93 made their first contribution in

  • @RafalSkolasinski made their first contribution in

  • @dumaas made their first contribution in

  • @Salehbigdeli made their first contribution in

  • @regen100 made their first contribution in

Full Changelog:…1.2.0


v1.2.0.dev1 - 01 Aug 2022


v1.1.0 - 01 Aug 2022