Skip to main content
​
DatasetsGraphics evaluatorsPythonSDKTypeScript

​
Introducing Graphics evaluators

Released Graphics evaluators to check the graphics in educational content. They take images as input, along with text:

​
Added Graphics Complexity dataset

Added a Graphics Complexity dataset that includes passages with graphics (grades 4-12), as well as expert-reviewed ratings for how difficult the graphics are to read.This dataset was used to validate the Graphics Complexity evaluator.

​
TypeScript SDK v1.2.0

Added support for the Graphics Accuracy and Graphics Complexity evaluators.

​
Python SDK v1.1.0

Added support for the Graphics Accuracy and Graphics Complexity evaluators.
​
PythonSDK

​
Introducing Python SDK v1.0.0

Added support for more Text Complexity and all Feedback and Academic Standards evaluators.This is a breaking release. There are no aliases for the old names.
  • Changed the evaluate() result to the evaluator / result / metadata envelope, replacing answer / explanation / metadata
  • Replaced the create_config* factories and the nested provider-config objects with one flat EvaluatorConfig. Pass that object, or pass the same fields as keyword arguments
    • Renamed telemetry_partner_id to learning_commons_api_key for Knowledge Graph and other Learning Commons APIs
    • Added max_retries, and a telemetry block that is on and anonymous by default. Telemetry never carries the text being evaluated. It has enabled and learning_commons_api_key only
    • Replaced per-step prompt_settings_step_*, default_evaluation_settings, and per-call evaluation_settings= with a single model_override, which requires both provider and model
    • Removed logger as a config field. Configure Python’s logging module instead
  • Rewrote the error taxonomy. APIError is now the abstract DependencyError, carrying dependency, status_code, request_id, and model. OutputValidationError is now LLMOutputProcessingError. RateLimitError.retry_after (seconds) is now retry_after_ms (milliseconds)
  • Renamed the grade input field to grade_level, and it is an int

​
Renames

Old ids still resolve for lookup. Results carry only the new id.

​
Default model changes

An evaluator running on a different model can return a different result for the same input.Vocabulary Complexity’s other steps are unchanged: background knowledge stays gpt-4o-2024-11-20, and the grades 3–4 step stays gemini-2.5-pro. To keep Grade Level Appropriateness on the v0.2.0 model, pin it with model_override.
​
PythonSDKTypeScript

​
Python SDK v0.2.1

  • Minor documentation updates

​
TypeScript SDK v1.1.1

  • Minor documentation updates
​
SDKTypeScript

​
TypeScript SDK v1.1.0

  • Minor internal refactoring
​
Durable Skills evaluators

​
Introducing Durable Skills evaluators

Released Durable Skills evaluators to rate student work against a construct’s indicators, with evidence and reasoning for every judgment:
​
SDKTypeScript

​
Introducing TypeScript SDK v1.0.0

Added support for more Text Complexity and all Feedback and Academic Standards evaluators.
  • Changed the evaluate() method’s inputs and outputs signatures
  • Updated configuration options
    • Renamed partnerKey to learningCommonsApiKey for Knowledge Graph and other Learning Commons APIs
    • Added llmProvider to inject your own LLM
    • Require both provider and model for modelOverride
  • Revised telemetry tracking behavior - On by default; anonymous unless you set learningCommonsApiKey on the telemetry option
  • Updated error types and made them all retryable
​
Text Complexity evaluatorsFeedback evaluators

​
Added Text Complexity evaluator

​
Added Feedback evaluators

​
Academic Standards evaluatorsSDKTypeScript

​
Introducing Academic Standards evaluators

Released Academic Standards evaluators to measure how aligned educational content is to U.S. academic standards:

​
TypeScript SDK v0.8.0

Added support for Math Standards Alignment.
​
Feedback evaluatorsText Complexity evaluators

​
Introducing Feedback evaluators

Released Feedback evaluators to assess the quality of AI-generated or teacher feedback:

​
Added Text Complexity evaluator

​
PythonSDKTypeScript

​
Python SDK v0.2.0

Added support for the Grade Level Appropriateness evaluator.

​
TypeScript SDK v0.7.0

Added --help, --version, and other options to the batch evaluator.
​
SDKPythonTypeScript

​
Introducing the Python SDK

Released a Python SDK ↗ with support for the following Text Complexity evaluators:

​
TypeScript SDK v0.6.0

Added --bypass-row-limit option for the batch evaluator.
​
Text Complexity evaluatorsSDKTypeScript

​
Added Text Complexity evaluator

​
TypeScript SDK v0.5.0

​
SDKTypeScript

​
TypeScript SDK v0.4.0

Added a Batch evaluator.
​
Text Complexity evaluatorsSDKTypeScript

​
Added Text Complexity evaluator

​
TypeScript SDK v0.3.0

Added support for Meaning Directness.
​
Text Complexity evaluatorsSDKTypeScript

​
TypeScript SDK v0.2.0

Added support for Background Knowledge Demands.
​
SDKTypeScriptLearning Commons PlatformText Complexity evaluators

​
Introducing the TypeScript SDK (v0.1.0)

Released a TypeScript SDK ↗ with support for the following Text Complexity evaluators:

​
Introducing the Evaluators playground

Released the Evaluators playground ↗, an interactive environment for testing evaluators without setup or integration.

​
Added Text Complexity evaluator

​
Text Complexity evaluators

​
Expanded Vocabulary Complexity support

Expanded support for Vocabulary Complexity to include grades 5–12.
​
Text Complexity evaluators

​
Expanded Sentence Structure support

Expanded support for Sentence Structure to include grades 5–12.
​
Text Complexity evaluators

​
Introducing Text Complexity evaluators

Released Text Complexity evaluators:

Support

Report issues or ask questions about a recent release.