MLServer

An open source inference server for your machine learning models.

Overview

MLServer aims to provide an easy way to start serving your machine learning models through a REST and gRPC interface, fully compliant with KFServing's V2 Dataplane spec. Watch a quick video introducing the project here.

• Multi-model serving, letting users run multiple models within the same process.
• Ability to run inference in parallel for vertical scaling across multiple models through a pool of inference workers.
• Support for adaptive batching, to group inference requests together on the fly.
• Scalability with deployment in Kubernetes native frameworks, including Seldon Core and KServe (formerly known as KFServing), where MLServer is the core Python inference server used to serve machine learning models.
• Support for the standard V2 Inference Protocol on both the gRPC and REST flavours, which has been standardised and adopted by various model serving frameworks.

You can read more about the goals of this project on the initial design document.

Usage

You can install the mlserver package running:

pip install mlserver

Note that to use any of the optional inference runtimes, you'll need to install the relevant package. For example, to serve a scikit-learn model, you would need to install the mlserver-sklearn package:

pip install mlserver-sklearn

For further information on how to use MLServer, you can check any of the available examples.

Inference Runtimes

Inference runtimes allow you to define how your model should be used within MLServer. You can think of them as the backend glue between MLServer and your machine learning framework of choice. You can read more about inference runtimes in their documentation page.

Out of the box, MLServer comes with a set of pre-packaged runtimes which let you interact with a subset of common frameworks. This allows you to start serving models saved in these frameworks straight away. However, it's also possible to write custom runtimes.

Out of the box, MLServer provides support for:

Framework	Supported	Documentation
Scikit-Learn	✅	MLServer SKLearn
XGBoost	✅	MLServer XGBoost
Spark MLlib	✅	MLServer MLlib
LightGBM	✅	MLServer LightGBM
CatBoost	✅	MLServer CatBoost
ONNX	✅	MLServer ONNX
Tempo	✅	github.com/SeldonIO/tempo
MLflow	✅	MLServer MLflow
Alibi-Detect	✅	MLServer Alibi Detect
Alibi-Explain	✅	MLServer Alibi Explain
HuggingFace	✅	MLServer HuggingFace

Runtime Security Maintainer Note

MLServer enforces a trusted runtime implementation allowlist in mlserver/settings.py (ALLOWED_MODEL_IMPLEMENTATIONS) when resolving model implementations from model-settings.json and MLSERVER_MODEL_IMPLEMENTATION.

If you add a new core runtime implementation shipped by this repository, you must also:

1. Add the runtime import path to ALLOWED_MODEL_IMPLEMENTATIONS.
2. Add or update tests validating allowlist behavior.
3. Keep runtime docs/examples aligned with the implementation import path.

For third-party or project-specific custom runtimes, do not extend the global allowlist. Use the image-scoped runtime workflow with mlserver build:

Building Images with Custom Runtimes:

# PRODUCTION mode (production): Allowlist specific custom runtimes
mlserver build . -t my-image \
  --allow-runtime models.MyRuntime \
  --runtime-path models.py

# DEVELOPMENT mode (development): Allow any runtime
mlserver build . -t my-dev-image --dev

Runtime Security Modes

MLServer operates in one of two security modes for loading custom runtimes:

PRODUCTION Mode (Production):

• Enforced when a trusted runtimes allowlist file exists in the image
• Only explicitly allowlisted runtimes can be loaded
• Custom runtimes must be baked into the image with --allow-runtime and --runtime-path
• Provides strong security guarantees for production deployments

DEVELOPMENT Mode (Development):

• Active when no allowlist file exists (e.g., running mlserver start directly)
• Supports dynamic loading of custom runtimes directly from model folders
• Simply place your custom runtime .py file next to model-settings.json
• Convenient for rapid local development and testing
• WARNING: Should NEVER be used in production - allows arbitrary code execution

You can query the current security mode through the /v2/runtimes REST endpoint or RuntimeSecurity gRPC method. See the model-settings reference and custom runtimes guide for details.

MLServer is licensed under the Apache License, Version 2.0. However please note that software used in conjunction with, or alongside, MLServer may be licensed under different terms. For example, Alibi Detect and Alibi Explain are both licensed under the Business Source License 1.1. For more information about the legal terms of products that are used in conjunction with or alongside MLServer, please refer to their respective documentation.

Supported Python Versions

🔴 Unsupported

🟠 Deprecated: To be removed in a future version

🟢 Supported

🔵 Untested

Python Version	Status
3.7	🔴
3.8	🔴
3.9	🟢
3.10	🟢
3.11	🟢
3.12	🟢
3.13	🔴

Examples

To see MLServer in action, check out our full list of examples. You can find below a few selected examples showcasing how you can leverage MLServer to start serving your machine learning models.

• Serving a scikit-learn model
• Serving a xgboost model
• Serving a lightgbm model
• Serving a catboost model
• Serving an onnx model
• Serving a tempo pipeline
• Serving a custom model
• Serving an alibi-detect model
• Serving a HuggingFace model
• Multi-Model Serving with multiple frameworks
• Loading / unloading models from a model repository

Developer Guide

Versioning

Both the main mlserver package and the inference runtimes packages try to follow the same versioning schema. To bump the version across all of them, you can use the ./hack/update-version.sh script.

We generally keep the version as a placeholder for an upcoming version.

For example:

./hack/update-version.sh 0.2.0.dev1

Testing

To run all of the tests for MLServer and the runtimes, use:

make test

To run run tests for a single file, use something like:

tox -e py3 -- tests/batch_processing/test_rest.py