Skip to content

Models & Projects

Overview

The models command group retrieves information about models (also called projects in the context of LLM tracing). The projects command is an exact alias — every subcommand works the same way under either name.

# These are identical:
arize_toolkit models list
arize_toolkit projects list
Command Description Client Method
models list List all models in the space get_all_models
models get Get a model by name get_model
models volume Get prediction volume get_model_volume
models total-volume Get total prediction count get_total_volume
models performance Pull performance metrics over time get_performance_metric_over_time
models delete-data Delete data from a model delete_data

models list

arize_toolkit models list

Lists all models in the current space with ID, name, type, and creation date.

Example

$ arize_toolkit models list
                         Models
┌──────────┬────────────────────┬─────────────────┬────────────┐
│ id       │ name               │ modelType       │ createdAt  │
├──────────┼────────────────────┼─────────────────┼────────────┤
│ m1       │ fraud-detection-v3 │ classification  │ 2024-06-01 │
│ m2       │ chatbot-prod       │ generative      │ 2024-09-15 │
└──────────┴────────────────────┴─────────────────┴────────────┘

# JSON output with jq
arize_toolkit --json models list | jq '.[].name'

models get

arize_toolkit models get NAME

Retrieves detailed information for a single model.

Arguments

  • NAME — The model name.

Example

arize_toolkit models get "fraud-detection-v3"
arize_toolkit --json projects get "chatbot-prod"

models volume

arize_toolkit models volume NAME [--start-time TIME] [--end-time TIME]

Returns prediction volume for a model, broken down by environment.

Arguments

  • NAME — The model name.

Options

  • --start-time (optional) — Start time in ISO format (e.g. 2025-01-01T00:00:00). Defaults to 30 days ago.
  • --end-time (optional) — End time in ISO format. Defaults to now.

Example

arize_toolkit models volume "fraud-detection-v3" --start-time 2025-01-01
arize_toolkit --json models volume "fraud-detection-v3"

models total-volume

arize_toolkit models total-volume [--model-name NAME] [--model-id ID] [--start-time TIME] [--end-time TIME]

Returns the total prediction count as a single number.

Options

  • --model-name (optional) — Model name.
  • --model-id (optional) — Model ID.
  • --start-time / --end-time (optional) — Time range in ISO format.

Example

$ arize_toolkit models total-volume --model-name "fraud-detection-v3"
Total volume: 142857

$ arize_toolkit --json models total-volume --model-name "fraud-detection-v3"
{"total_volume": 142857}

models performance

arize_toolkit models performance METRIC ENVIRONMENT [OPTIONS]

Pulls a performance metric time-series for a model.

Arguments

  • METRIC — Performance metric name (e.g. accuracy, auc, f1).
  • ENVIRONMENT — Environment name (e.g. production, training).

Options

  • --model-name (optional) — Model name.
  • --model-id (optional) — Model ID.
  • --granularity — Time granularity: hour, day, week, month. Defaults to month.
  • --start-time / --end-time (optional) — Time range in ISO format.

Example

arize_toolkit models performance accuracy production --model-name "fraud-detection-v3" --granularity day
arize_toolkit --json models performance auc production --model-name "fraud-detection-v3"

models delete-data

arize_toolkit models delete-data NAME [--start-time TIME] [--end-time TIME] [--yes]

Deletes data from a model within the specified time range. This is a destructive action — the CLI will prompt for confirmation unless --yes is passed.

Arguments

  • NAME — The model name.

Options

  • --start-time / --end-time (optional) — Time range in ISO format.
  • --yes — Skip the confirmation prompt.

Example

arize_toolkit models delete-data "old-model" --start-time 2024-01-01 --end-time 2024-06-01