NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #4665 most downloaded on PyPI
A helper library to interact with Arize AI APIs
Last release today
17 Sep 2026
Ships on a steady schedule
a new release about every 2 weeks
Nearly every release is documented
notes for 58 of the last 60 stable releases
10 versions withdrawn
withdrawn after publishing
6 years old
364 releases · first in 2020
One column per quarter.
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Correctly insert default prediction id column using df.insert()
df.insert() (f938d29)python exporter embedding similarity search support
Serialization of nested dictionaries
Add session and user ids to spans batch logging
Add missing __init__.py file to tracing validation module
Add log_evaluations method for delayed evaluation logging
Increase embedding raw data character limit
### 🐛 Bug Fixes * Allow spaces in eval names
Support export of spans from Arize platform
use pandas items() vs deprecated iteritems()
sdk: Move version to version.py
Add certificate file reading to sdk client
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Fix ImportError when importing Client from arize.api
ImportError when importing Client from arize.apiAdd optional extra dependencies if the Arize package is installed as pip install arize[NLP_Metrics]:
pip install arize[NLP_Metrics]:
nltk>=3.0.0, <4sacrebleu>=2.3.1, <3rouge-score>=0.1.2, <1evaluate>=0.3, <1Nothing published for this version
Nothing published for this version
Nothing published for this version
Address backward compatibility issue for batch logging via Pandas for on-prem customers
Add deprecated to our Tracing extra requirements. The deprecated dependency comes from opentelemetry-semantic-conventions, which absence produced an I…
deprecated to our Tracing extra requirements. The deprecated dependency comes from opentelemetry-semantic-conventions, which absence produced an ImportErrorRelax MimicExplainer extra requirements: require only interpret-community[mimic]>=0.22.0,<1
MimicExplainer extra requirements: require only interpret-community[mimic]>=0.22.0,<1MULTICLASS model typeTRACING environment. You can now log spans & traces for your LLM applications into Arize using batch ingestion via Pandas DataFramesSchema. You can now log wider models (more columns in your DataFrame)Nothing published for this version
Nothing published for this version
Nothing published for this version
New MULTICLASS model type available for record-at-a-time ingestion
MULTICLASS model type available for record-at-a-time ingestionFix missing columns validation feedback to have repeated columns in the message
KeyError when llm_params is not found in the dataframe. Improved feedback to the user was included.Updated pandas requirement. We now accept pandas 2.x
pandas requirement. We now accept pandas 2.xGENERATIVE_LLM modelsDefault prediction sent as string for GENERATIVE_LLM single-record-logger (before it was incorrectly set as an integer, resulting in it being categori
GENERATIVE_LLM single-record-logger (before it was incorrectly set as an integer, resulting in it being categorized as prediction score instead of prediction label)Only check the value of prompt/response if not None
prompt/response if not NoneAccept strings for prompt and response
CORPUS supportNothing published for this version
Add validation on embedding raw data for batch and record-at-a-time loggers
Add ability to send features with type list[str]
Require python>=3.6 (as opposed to python>=3.8) for our core SDK. Our extras still require python>=3.8.
python>=3.6 (as opposed to python>=3.8) for our core SDK. Our extras still require python>=3.8.pyarrow>=0.15.0 (as opposed to pyarrow>=5.0.0)Add prompt templates and LLM config fields to the single log and pandas batch ingestion. These fields are used in the Arize Prompt Template Playground
Nothing published for this version
Add filtering via the keyword where to the Exporter client
Computer Vision AutoEmbeddings switched from using FeatureExtractor(deprecated from HuggingFace) to ImageProcessor class
AutoEmbeddings support for any model in the HuggingFace Hub, public or private.AutoEmbeddings UseCase for Object DetectionEmbeddingGenerator.list_default_models() methodAutoEmbeddings switched from using FeatureExtractor(deprecated from HuggingFace) to ImageProcessor classAuthenticating Arize Client using environment variables
Add Generative_LLM model-type support for single-record logging
Generative_LLM model-type support for single-record loggingRemoved dependency on interpret for the MimicExplainer
Add missing dependency for Exporter: tqdm>=4.60.0,<5
GENERATIVE_LLM model prompt & response fieldsNothing published for this version
Relax protobuf requirements from protobuf~=3.12 to protobuf>=3.12, <5
protobuf~=3.12 to protobuf>=3.12, <5Add new ExportClient, you can now export data from Arize using the Python SDK
ExportClient, you can now export data from Arize using the Python SDKREGRESSION models to use the MimicExplainerprediction_label and actual_label from single-record loggingOBJECT_DETECTION modelsNothing published for this version
Change optional dependency for MimicExplainer, raise the version ceiling of lightgbm from 3.3.4 to 4
MimicExplainer, raise the version ceiling of lightgbm from 3.3.4 to 4NUMERIC model typesNothing published for this version
Fix GENERATIVE_LLM models being sent as SCORE_CATEGORICAL models
GENERATIVE_LLM models being sent as SCORE_CATEGORICAL modelsNothing published for this version
Require Python >= 3.8 for all extra functionality
Python >= 3.8 for all extra functionalitynumeric_sequence supportAdd optional extra dependencies if the Arize package is installed as pip install arize[LLM_Evaluation]:
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 →