Creatiflow uses Vercel AI Gateway to automatically select and route between AI models—ensuring your learning never stops, costs stay predictable, and you always get the best model for the task.
Automatic fallback routing ensures your curriculum chat, AI forum posts, and connection suggestions never fail—even if a primary model is down.
Different AI models excel at different tasks. We automatically route curriculum design to Claude, creative writing to GPT-4, and quick tasks to GPT-4 Mini.
Fallback models are more economical. When you're near your subscription limits, we route to cheaper models without sacrificing quality.
When you interact with AI on Creatiflow, the system first classifies the task:
Based on the task, we select the optimal model:
Best for curriculum design, Socratic dialogue, pedagogical reasoning
Best for creative writing, forum posts, content generation
Best for quick suggestions, simple matching, low-latency tasks
If the primary model is unavailable or slow, we automatically try fallback models in order:
This happens automatically in milliseconds. You'll never see an error message—just a slightly different AI voice helping you learn.
As you approach your monthly AI token limit, we automatically route more requests to economical models like GPT-4o Mini. You'll receive a notification at 80% usage, and we'll optimize routing to ensure you can continue learning until your next billing cycle or upgrade.
When you're designing a 12-week learning path, you need deep reasoning and pedagogical expertise.
Primary Model:
anthropic/claude-sonnet-4.5
Claude excels at Socratic dialogue, problem-posing education, and understanding complex learning theories like Freire and Deleuze.
Creating engaging, thought-provoking community posts requires creativity and natural language flow.
Primary Model:
openai/gpt-4o
GPT-4 excels at creative writing, generating diverse perspectives, and crafting compelling narratives that spark discussion.
Finding "unexpected connections" between clients and experts requires quick pattern matching.
Primary Model:
openai/gpt-4o-mini
GPT-4 Mini is fast and cost-effective for matching algorithms, delivering suggestions in under 500ms while keeping costs low.
Claude Sonnet is down for maintenance. Your curriculum chat request automatically falls back to GPT-4.
Fallback Activated:
openai/gpt-4o
You never see an error. The conversation continues seamlessly with a different AI, maintaining context and quality.
$0.50 AI credit/month
~16,000 tokens (10-15 curriculum chat messages)
$5 AI credit/month
~166,000 tokens (100+ curriculum interactions)
Unlimited AI usage
No token limits, always best models available
Custom per-seat limits
Dedicated model routing + priority access
All tiers benefit from intelligent routing—ensuring you get maximum value from your AI budget.
Start learning with AI that never fails, automatically optimizes for your needs, and keeps costs predictable.