PromptUI gives you access to 11 AI models across 3 tiers. The right choice depends on what you're building, how fast you want it, and how much quality you need.
Fast Tier (5–10 credits)
Best for: Quick prototypes, simple landing pages, testing ideas.
Aurora Alpha (5 credits) — Free reasoning model. Great for trying PromptUI out, but struggles with complex multi-file apps.
Grok 4.1 Fast (10 credits) — Ultra-fast with a 2M context window. Good for dashboards and data-heavy apps. Occasionally produces preview errors.
DeepSeek V3.2 (10 credits) — GPT-5 class reasoning at a fraction of the cost. Strong on CRUD apps and backend logic. Some reliability variance.
MiniMax M2.5 (10 credits) — #1 on Programming benchmarks this week (80.2% SWE-Bench). Worth trying for code-heavy projects.
Balanced Tier (10–15 credits)
Best for: Production apps, multi-page projects, anything you plan to share.
Kimi K2.5 (10 credits) — #2 on Programming benchmarks. Strong multimodal and tool-calling, 262K context. Our pick for the price-conscious builder who still wants quality.
Gemini 3 Flash (10 credits) — Near-Pro reasoning with a massive 1M context window. Good for large apps, experimental reliability.
Claude Sonnet 4.5 (15 credits) — Previous gen Sonnet. Still excellent, 1M context. If Sonnet 4.6 ever has issues, fall back here.
Claude Sonnet 4.6 (15 credits) — Our default and recommended model. #1 adoption growth (+363%), frontier-tier coding. Produces the cleanest, most reliable component structures across all app types. Use this unless you have a specific reason not to.
Powerful Tier (25 credits)
Best for: Complex SaaS apps, investor demos, portfolio pieces where quality is everything.
Claude Opus 4.6 (25 credits) — The most capable model available. #4 on Programming benchmarks, 1M context, best for large codebases and complex business logic. Use when you want the absolute best output and credits aren't the primary concern.
Quick Reference
| Use case | Recommended model | Credits |
|---|---|---|
| Quick prototype / testing | Aurora Alpha | 5 |
| Dashboard / data app | Grok 4.1 Fast | 10 |
| Standard app / multi-page | Claude Sonnet 4.6 | 15 |
| Complex SaaS / investor demo | Claude Opus 4.6 | 25 |
| Budget-conscious quality | Kimi K2.5 | 10 |
All models are available on every plan — the only limit is credits. Upgrade or buy top-ups any time from your dashboard.