NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #5026 most downloaded on PyPI
A helper library to interact with Arize AI APIs
Last release 5 days ago
29 Sep 2026
Ships on a steady schedule
a new release about every 2 weeks
Nearly every release is documented
notes for 59 of the last 60 stable releases
10 versions withdrawn
withdrawn after publishing
7 years old
367 releases · first in 2020
One column per quarter.
Support Sending in Additional Columns for Experiments
Update evals logging to allow session_eval. and trace_eval. prefixes in addition to eval.
delete_experiment dependency sync
experiments: Add delete experiment method to ArizeDatasetsClient
add log_metadata method for LLM span metadata updates
New README landing page for arize package
Update prompt hub client to use unified params for push prompt
pandas logger Client, ArizePromptClient(experimental): deprecate developer_key in favor of api_key in pandas logger Client and ArizePromptClient const…
experimental, datasets: deprecate developer_key parameter in ArizeDatasetsClient
### 🎁 New Features * log_annotations
Add and mark underscored headers for deprecation
### 🎁 New Features * CV: Image segmentation support
experimental, prompt-hub: Prompt Hub Client
experimental, whylabs-vanguard: add log_dataset_profile env parameter
experimental, online_tasks: Add extract_nested_data_to_column for preprocessing
experimental, whylabs: profile view support
experimental, whylabs-vanguard: add graphql endpoint parameter, default model env for governance
experimental, whylabs: fix timestamp logic
experimental, whylabs-vanguard: fix profile adapter (generator) name
experimental, whylabs: integration client implementing log_profile, log_dataset
experimental, whylabs: Rename adapter class
(experimental, whylabs) Remove implementation of Synth Data Generation from profiles
(experimental): synthetic data generator for whylabs integration
strip scheme from host in exporter client argument
Use the correct type for open inference attributes.llm.tools column
Add support for Open Inference llm.tools attribute
handle pd.NA type for surrogate explainability
Add obj parameter to isinstance in _datetime_to_ns
Fix tool calls function arguments validation for output_messages when logging spans as a dataframe
fix missing classes in api-reference
datasets: log experiment dataframe
Increase max character length for multiclass class name to 100
Update switcher.json for readthedocs
allow protobuf v5 for readthedocs doc build
new log_evaluations_sync method to public Flight server
datasets: default experiment traces for the same dataset are sent to Arize in 1 traces model/project
relax googleapis-common-protos and protobuf dependencies
limit dataset identifier hash to 8 characters
Add version switcher dropdown to api docs
Add switcher to conf.py in api docs
### 📚 Documentation * Add versions to switcher.json
Add mps as available device for embedding generator
### ❔ Miscellaneous Chores * fix path for readthedocs.yaml
Exclude api_reference directory from build
Move arize files into src directory
datasets: support using dataset_name in addition to dataset_id to update/delete/get_versions on a dataset
Adds project_name support to the Tracing SDK, alongside, to replace model_id, which will be deprecated for Tracing
experiment: evaluation result out of order
Improve export of arize tracing data to improve ergonomics of working with data after export
### 📚 Documentation * update readme
### 📚 Documentation * minot update README.md
### 📚 Documentation * Update README.md
experiments: configurable concurrency value for task and evaluator runs
async run task and evaluator functions
Nothing published for this version
Nothing published for this version
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →