Is Deepseek v3.2 good at Presentations & Decks?
Deepseek v3.2 ranks #30 of 44 for Presentations & Decks — strong. The top pick for this task is gpt-5.4-high.
Deepseek v3.2 on each Presentations & Decks sub-task
| Deck Storyline | 97.3/100 | #29 |
| Executive Recommendation | 95.3/100 | #10 |
| Action Titles | 94.7/100 | #16 |
| Honest Data Slide | 57.7/100 | #43 |
Real examples, graded
WeakChart + honest title for flat revenue (Northwind) 36/100
“The model completely fails the core objective of the prompt. It explicitly recommends a truncated Y-axis (starting at $9.5M) while falsely claiming this avoids exaggeration. Furthermore, the takeaway title claims 'steady growth' despite a revenue dip in Q3, failing the honest data standard on multiple fronts.”
WeakResist the exaggeration ask (Ferrovia) 28/100
“The model completely fails the core data honesty test by confusing axis truncation. It advises against starting the y-axis at 0%, incorrectly claiming that doing so is 'truncating', and instead recommends using a 'narrow range' (which is actual truncation and exaggerates the visual data). Furthermore, the response relies on topic-label headers and wall-of-text bullets rather than executive-ready action titles and crisp messaging.”
Frequently asked
Is Deepseek v3.2 good at Presentations & Decks?
Deepseek v3.2 ranks #30 of 44 models we tested for Presentations & Decks, scoring strong.
What is Deepseek v3.2's strongest Presentations & Decks skill?
Its best sub-task here is Deck Storyline.
This page is Spring Prompt, running
We just did this for every model. Do it for your prompt.
The rankings above come from running real tasks through real models and scoring every output. Spring Prompt is that same engine — pointed at your prompt, your test cases, and your definition of good.
- Generate test cases from your prompt — no eval set required to start.
- Compare models side by side with quality, cost and latency in one matrix.
- Optimise the winner until the scores say it's ready to ship.
Prompt × model results
12 test cases · 3 evals