Choosing the Foundry Model for Generative Text Suggestions

AI-103 By islever Editorial Last updated

Choose the appropriate Foundry services for generative AI and agents Build generative applications by using Foundry

Short answer

D Use GPT-4, a generative chat model, to turn the product description prompt into newly generated product-name text in Microsoft Foundry.

Question

You have an Azure subscription that contains Microsoft Foundry resource.
You need to build an app that will suggest product names from a given product description.
Which Foundry model should you use?
  1. Whisper
  2. DALL-E
  3. embeddings
  4. GPT-4 Correct Answer

Key Takeaways

The trap is matching the task's output modality to the model: naming requires new text tokens, while Whisper returns transcripts, DALL-E returns images, and embeddings return only numeric vectors.

This AI-103 question tests model selection in Microsoft Foundry: suggesting product names from a description is a text-generation task, so the correct model is GPT-4 (D), not Whisper, DALL-E, or embeddings.

Common mistake: Choosing C (embeddings), because embeddings are heavily associated with Foundry grounding and retrieval; but an embedding model outputs floating-point vectors with no readable product names.

Expert Analysis

What the app actually does. The requirement is to take a natural-language product description and produce new text — a list of creative, plausible product names. That is a text-generation task, not transcription, image rendering, or vector encoding. Every request therefore has to end at a model that emits tokens autoregressively from a prompt.

Tracing the request through Foundry. The app would send the description in a messages array to a chat-completions endpoint in your Foundry resource, typically with a system prompt such as "Suggest five product names and a one-line rationale." The model conditions on the description plus the instruction and writes fresh output tokens. This is exactly the chat/complete capability exposed by GPT-class models, so option D (GPT-4) is the correct choice.

Why the other options fail at different steps. Option A (Whisper) is a speech-to-text model: its input path expects audio, and it transcribes rather than invents, so it never reaches a naming step. Option B (DALL-E) is an image-generation model whose output is pixels, not names — it would illustrate the product, not name it. Option C (embeddings) maps text to a numerical vector; embeddings are used for semantic search, clustering and retrieval grounding, and they produce no readable text at all. An embeddings model could help later if you wanted to rank candidate names against existing catalogue items, but it cannot be the model that suggests the names themselves.

Trap awareness. One comment correctly marks the answer as D, yet its body then explains embedding models and vectors — a mismatch between the letter and the justification. Read the explanation, not just the letter: converting a description into [0.0142, -0.0284,...] is precisely what option C does, and it would leave the app with nothing to display. The decisive test on AI-103 is always: what is the output modality? Text in, new text out = a GPT chat/completion model.

Official Reference

https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/models

This link opens the official vendor documentation in a new tab.

Exam Strategy

For every "which model should you use" question, first write down the required output modality: spoken-text transcript, image, vector, or generated text. Then eliminate options whose output type does not match before judging any feature list. On AI-103, embeddings are the most frequent distractor — if the task needs human-readable output, an embedding model can never be the primary answer, even if retrieval is involved elsewhere in the design.

Frequently Asked Questions

Why can't an embeddings model suggest product names in Foundry?

An embeddings model converts the description into a vector of floating-point numbers for search and similarity; it emits no readable text, so there is nothing to display as a product name. Use embeddings only to rank or retrieve candidates afterwards.

When would Whisper or DALL-E apply in a product-naming app?

Whisper applies only if the description arrives as audio that must be transcribed first. DALL-E applies if you also need a product image or mock-up; neither can generate the written names themselves.

Could I use embeddings together with GPT-4 for product naming?

Yes, as a secondary step: embed existing catalogue names and the generated candidates, then use vector similarity to filter duplicates or rank novelty. The naming itself must still be done by a GPT-4 chat model.

More AI-103 FAQ →

Community Votes

D
100%

100% of anonymous learners picked answer D. Votes are pick records left by other test-takers — they are not the verified answer.

Discussion 1 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.

From public community threads 1

  1. 8ac3742 👍 1 Answered D
    An embedding model (such as text-embedding-3-small or text-embedding-ada-002) converts text into a mathematical vector—a list of floating-point numbers:$$\text{"A lightweight running shoe with breathable mesh"} \longrightarrow [0.0142, -0.0284, 0.0891, \dots]$$

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Which Request Parameters Maximize Foundry Output Stability and Reasoning?

S-grade

Azure Custom Vision: ordered steps to detect factory component faults

S-grade

Improving Agent Response Completeness in Microsoft Foundry

S-grade

Restricting Access to Microsoft Foundry Service via VNet

S-grade

Practice All AI-103 Questions

Access 88 questions with complete answers and detailed explanations.

View Full AI-103 Practice Test →

← Back to AI-103 Study Guide