Editing · How-to
How to regenerate a slide with a different AI model
Not happy with one slide? Regenerate just that slide with any of seven AI models — Llama 3.3 70B, Llama 3.1 8B, Llama 4 Scout, Qwen 3, or GPT-OSS 20B/120B. The whole slide (bullets, tables, and data charts) is rewritten with fresh, factual content, leaving the rest of your deck untouched.
Step by step
- 1Open your deck in the editorGenerate or open a deck. Each slide shows in the left thumbnail rail and in the main preview.
- 2Open the regenerate menuRight-click any slide in the left rail and choose “Regenerate with model”, or use the “Regenerate slide” button just below the slide preview.
- 3Pick a modelChoose from the list — each shows its provider and credit rate. Llama 3.3 70B is the balanced default; Llama 4 Scout is fastest with the most headroom; GPT-OSS and Qwen offer higher-throughput alternatives.
- 4Review the resultThe slide's title, bullets, tables, and any data chart are rewritten with fresh, real data. If it's a chart slide, the graph itself is replaced — not just the text.
- 5Repeat per slide as neededRegenerate is per-slide, so you can mix models across a deck and keep the versions you like. Credits are charged by tokens used × the model's rate.
Try it now
It takes seconds and the free plan needs no card.
Open the editorFrequently asked questions
Does regenerating change my whole deck?
No — it only rewrites the one slide you choose. Your other slides, theme, and template stay exactly as they are.
Which model should I use?
Llama 3.3 70B (the default) is the most reliable all-rounder. Llama 4 Scout is fast with lots of token headroom. See the /benchmarks page for measured speed and reliability of every model.
Does it change the charts too?
Yes. On a slide with a data chart, regeneration replaces the chart's data with different, real figures — the rendered graph updates in place, not only the surrounding text.
How much does it cost?
Regeneration is metered by the tokens used multiplied by the chosen model's credit rate, so a longer slide or a pricier model costs more. Rates are shown in the model menu and on the benchmarks page.