3.2 — Model options & cost
On this page
3.2 — Model options & cost
VibeMap uses frontier models (OpenAI, Anthropic, Google) under the hood. You can usually leave the defaults alone — but here's what's happening if you want to know.
Default model behavior
By default, VibeMap picks the right model for each generation type automatically:
- Quick generations (summary, persona suggestions) use fast, cheap models.
- Heavy generations (acceptance criteria for a large feature set, sprint planning) use stronger models.
- Fallbacks kick in automatically — if one provider is down, VibeMap retries with another. You won't notice unless everything is down.
For 95% of users this is exactly right.
Overriding the model
On most generation triggers, you can pick a model manually:
- Fast — for quick iteration. Cheapest, lowest latency, slightly less rigorous output.
- Balanced (default) — what we recommend for almost everything.
- Deep — for complex acceptance criteria or schema. Slower and more expensive but markedly better at long, structured output.
📸 Placeholder:
16-model-selector.png— the model selector dropdown on a generation button.
What generation costs you
Generations are metered against your plan's monthly allowance. The free tier covers a handful of full project generations and many small regenerations per month. The Pro tier covers serious daily use.
You can see your current usage on the Billing page.
📸 Placeholder:
17-billing-usage.png— the Billing page showing usage against allowance.
Tips for keeping cost down
- Regenerate parts, not the whole project. If only your acceptance criteria are off, regenerate just those — don't rerun the whole pipeline.
- Use Fast for first drafts, Deep only when you need it. Most generations are perfectly usable from the Fast tier.
- Edit before regenerating. A 30-second edit often beats a 3-minute regenerate-and-hope.