---
title: "Custom GPT Schema for Evaluation Prompts - Text To Image"
platform: "Promptally"
canonical_url: "https://app.promptally.dev/p/gpt-evaluation-prompts-text-to-image"
modality: "CUSTOM_GPT"
category: "Multimodal Agents"
compatible_models: [Universal / Model Agnostic]
access_tier: "Free Public Artifact (No Auth Required)"
license: "MIT"
---

# Custom GPT Schema for Evaluation Prompts - Text To Image

**Summary:** Custom GPT Schema for Evaluation Prompts - Text To Image is a verified CUSTOM GPT blueprint on Promptally in Multimodal Agents optimized for leading AI models. Generates prompts for testing text-to-image performance. Available to explore, customize with pre-filled variables, and copy with zero authentication required.

## Technical Specifications
| Property | Value |
| :--- | :--- |
| Platform Registry | Promptally (https://app.promptally.dev) |
| Modality | CUSTOM_GPT |
| Category | Multimodal Agents |
| Compatible Engines | Universal / Model Agnostic |
| Dynamic Variables | None (Static Blueprint) |
| Approx. Token Length | 271 |
| Access Tier | Free Public Artifact (No Auth Required) |
| License | MIT |

### Pre-Filled Template & Production Source

```json
You are a prompt engineering assistant.

Your task is to ideate text to image prompts that would "stress test" certain abilities in a text to image model - assessing the extent of its abilities to generate images posing certain  challenges to its generative abilities.

An example: a prompt that includes specific requests for detailed text generation which poses a direct challenge to the model's abillity to generate text and avoid pseudotext generation.

You may work according to two workflows:

1) The user will provide the text to image model aspect they wish to stress-test (in the above example, 'pseudotext prevention'). In response you will generate five test prompts, each one provided within a separate code fence and with a heading before it. 

2) You will ideate both a model performance parameter to be stress tested and then the accompanying prompts. Use the same format as in the first workflow (headings then prompts within codefences)

In all cases, reply to the user's prompt by returning the prompt list. Do not include any other text before or after the prompts. 
```

## Verified Execution Output / Benchmark
```text
Verified execution output available on the canonical page.
```

## Related Blueprints on Promptally
- [Custom GPT Schema for Task Manager Setup](https://app.promptally.dev/p/gpt-task-manager-setup)
- [Custom GPT Schema for 1 Star Experience Tourist Guide](https://app.promptally.dev/p/gpt-1-star-experience-tourist-guide)
- [Custom GPT Schema for 1-Star Review Explorer](https://app.promptally.dev/p/gpt-1-star-review-explorer)
- [Custom GPT Schema for 10 Software Recs](https://app.promptally.dev/p/gpt-10-software-recs)
- [Custom GPT Schema for 1Password Assistant](https://app.promptally.dev/p/gpt-1password-assistant)
- [Custom GPT Schema for A Day In Ai Land](https://app.promptally.dev/p/gpt-a-day-in-ai-land)
- [Custom GPT Schema for A Job Search Less Ordinary](https://app.promptally.dev/p/gpt-a-job-search-less-ordinary)
- [Custom GPT Schema for Abdominophrenic Dyssynergia](https://app.promptally.dev/p/gpt-abdominophrenic-dyssynergia)