Graphics Complexity

Evaluator last updated October 14, 2026.

Overview

The Graphics Complexity evaluator assesses how much complexity the graphics add to a text’s reading complexity. A graphic might be a decoration a reader can skip, or the only place a key idea appears. The evaluator works in 2 steps:

Identifies each graphic’s role: Determines how much meaning each graphic carries relative to the prose.
Evaluates how hard it is to read: Weighs what each graphic takes to read (density, layering, interpretive elements like legends or dual axes, and abstraction) against what students at the target grade are expected to handle independently. When a text has more than one graphic, the score reflects the most demanding graphic, or the combined demand of graphics that must be read together.

At a glance


Supported grades	3–12
Rubric	Student Achievement Partners’ Qualitative Text Complexity Rubric for Informational Text (Use of Graphics)
Model used	Gemini 3 Flash preview (`gemini-3-flash-preview`) with temperature of 1 The evaluator was built and validated using this specific model and temperature. Other configurations will produce different results and may have lower accuracy.

Getting started

Follow the Quickstart to start using this evaluator:

Access method
SDK	Python ↗ (v1.1.0 or later) and TypeScript ↗ (v1.2.0 or later)
Other	Python notebook ↗ Prompts: `system.txt` ↗ and `user.txt` ↗

Inputs

Requirement	Supported	Required
Text	Informational text that the graphics accompany 1–10,000 characters	Yes
Graphics	1–5 image files, one per graphic, in reading order PNG, JPEG, or WebP, up to 2.5 MB each. Width and height must each be 16–8,000 pixels. The format is read from the file’s contents, not its extension. An image outside these limits is rejected, not resized. Animated graphics aren’t supported. For an interactive graphic, submit a screenshot of the view that the text discusses.	Yes
Grade level	Target student grade level	Yes
Figure labels	Labels for the graphics, comma-separated, in the same order as the images (e.g., “Figure 1, Figure 2”). If you leave this out, graphics are labeled “Image 1,” “Image 2,” and so on	No

Submit each graphic the way a reader would see it: the full frame, with labels and legends legible. A cropped or low-resolution image can change the score.

Captions outside the image aren’t evaluated. Labels, axes, keys, and legends inside the image are.
Each image counts as one graphic, even if it has several panels.
If you’re pulling figures from a PDF, render them at 200 dpi, the resolution used when building the evaluator.

Output

Field	Description
Complexity score	Graphics complexity level: Slightly complex: Simple and directly illustrative, or not needed to understand the text (e.g., a photograph that supports the prose) Moderately complex: Clear, but needs limited integration with the text and may include interpretive elements (e.g., a flowchart with a color key); understanding is deeper with the graphic, but not blocked without it Very complex: Must be integrated with the text, or requires moving between multiple layers of information (e.g., a heatmap that coordinates time, altitude, and density) Exceedingly complex: Requires moving flexibly between multiple layers or types of information, and presents dense information that isn’t in the text More context needed: A graphic is missing, doesn’t render, or is illegible
Reasoning	High-level summary of why the text’s graphics are at this complexity level for the target grade, including which graphic drives the rating and whether it’s the graphic itself or its role relative to the prose
Details	Individual complexity factors driving the rating, how they manifest in the graphics, and how they affect the text’s graphics complexity

Interpreting results

Output	How to use
Complexity score	Slightly complex: Confirm the graphics won’t get in the way of reading the prose. If the text is meant to build students’ skill with graphics, consider a graphic that carries more meaning Moderately complex: Point students to the graphic, since it deepens understanding even though the text stands on its own Very complex: Check that students at the target grade have learned to read this kind of graphic, and plan scaffolding or simplify it. The evaluator tends to stop at Very complex where experts would rate a graphic higher, so review Very complex data graphics closely Exceedingly complex: Decide whether the demand is deliberate. Key content likely appears only in a graphic students may not be able to read yet More context needed: Resubmit the graphic as a full-frame, legible image
Reasoning	Find which graphic drives the rating, and whether the cause is what it takes to read or its role relative to the prose. Then regenerate the graphic, restate its content in the prose, simplify the display, or keep the complexity as a deliberate choice. Rerun the evaluator to confirm the rating changed
Details	Use the scaffolding suggestions to support students with demanding graphics Example: Direct students to the graphic before the prose, walk through how to read that type of display, or ask what the graphic does and doesn’t show

Evaluator release history

Date	Changed
October 14, 2026	First release

Graphics evaluators

Explore other evaluators that assess the graphics in a text.

Text Complexity evaluators

Explore evaluators that assess the qualitative complexity of a text’s prose.

Quickstart

Run an evaluator in the Evaluators playground, a Python notebook, or with the SDK.

Understanding evaluators

Getting started

SDK API reference

Text Complexity evaluators

Graphics evaluators

Feedback evaluators

Academic standards evaluators

Durable skills evaluators

Datasets

Resources

Overview

At a glance

Getting started

Inputs

Output

Interpreting results

Evaluator release history

Graphics evaluators

Text Complexity evaluators

Quickstart

​Overview

​At a glance

​Getting started

​Inputs

​Output

​Interpreting results

​Evaluator release history

​Related topics

Graphics evaluators

Text Complexity evaluators

Quickstart

Overview

At a glance

Getting started

Inputs

Output

Interpreting results

Evaluator release history

Related topics