Documentation Index
Fetch the complete documentation index at: https://arize-ax.mintlify.dev/docs/llms.txt
Use this file to discover all available pages before exploring further.
This guide helps you migrate from Arize Python SDK v7 to v8.
Key Architectural Changes
Unified Client
The new v8 introduces a single ArizeClient that replaces all specialized clients from v7.
This provides a unified, discoverable interface following the pattern:
client.<resource>.<action>()
space_id explicitly per call, rather than
configuring it once during client initialization. This enables working with
multiple spaces from a single client instance.
Keyword-only arguments
The new v8 methods use keyword-only parameters (using * in signatures) to improve code
clarity and prevent positional argument errors.
Quick Reference
This section provides a quick lookup table for migrating from v7 to v8 methods. Each subsection shows the client initialization change and method mappings. For detailed migration guides with parameter changes and code examples, see the linked pages.
Pandas Client
The Pandas Client is used for batch logging of LLM traces, evaluations, and traditional ML model predictions using pandas DataFrames.
from arize.pandas.logger import Client
client = Client(...)
| v7 | v8 | Details |
|---|
client.log() | client.ml.log() | See details → |
client.log_spans() | client.spans.log() | See details → |
client.log_evaluations() | client.spans.update_evaluations() | See details → |
client.log_evaluations_sync() | client.spans.update_evaluations() | See details → |
client.log_annotations() | client.spans.update_annotations() | See details → |
client.log_metadata() | client.spans.update_metadata() | See details → |
Stream Client
The Stream Client is used for real-time logging of model predictions, providing lower latency than batch logging.
from arize.api import Client
client = Client(...)
| v7 | v8 | Details |
|---|
client.log() | client.ml.log_stream() | See details → |
Datasets & Experiments
The Datasets & Experiments Client is used for creating and managing datasets, running experiments, and tracking evaluation results.
from arize.experimental.datasets import ArizeDatasetsClient
client = ArizeDatasetsClient(...)
Datasets
| v7 | v8 | Details |
|---|
client.list_datasets() | client.datasets.list() | See details → |
client.create_dataset() | client.datasets.create() | See details → |
client.get_dataset() | client.datasets.get() + client.datasets.list_examples() | v8 splits into metadata (datasets.get()) and examples (datasets.list_examples()). See details → |
client.update_dataset() | N/A | Coming soon in v8 |
client.delete_dataset() | client.datasets.delete() | See details → |
client.get_dataset_versions() | N/A | Merged into get() response |
Experiments
| v7 | v8 | Details |
|---|
client.run_experiment() | client.experiments.run() | See details → |
client.log_experiment() | client.experiments.create() | See details → |
client.get_experiment() | client.experiments.get() + client.experiments.list_runs() | v8 splits into metadata (experiments.get()) and runs data (experiments.list_runs()). See details → |
client.delete_experiment() | client.experiments.delete() | See details → |
Exporter
The Exporter Client is used for exporting data from Arize back to pandas DataFrames or Parquet files for analysis, debugging, or building custom workflows.
from arize.exporter import ArizeExportClient
client = ArizeExportClient(...)
| v7 | v8 | Details |
|---|
client.export_model_to_df() (for spans/LLM) | client.spans.export_to_df() | See details → |
client.export_model_to_parquet() (for spans/LLM) | client.spans.export_to_parquet() | See details → |
| | |
client.export_model_to_df() (for models) | client.ml.export_to_df() | See details → |
client.export_model_to_parquet() (for models) | client.ml.export_to_parquet() | See details → |
Next Steps
For detailed migration guides with code examples:
