Choose a route for the task
A transparent guide to SmaxAI's routing profiles, verified creative studios, and supported foundation models.
Pragg Air — agile routing
Use Pragg Air for everyday prompts when you prefer a speed-oriented configured provider cascade. It dynamically balances through high-throughput pools (Groq, Cerebras, NVIDIA NIM) with fallback to Gemini Flash, delivering sub-second response times.
Pragg Max — reasoning cascade
Use Pragg Max for tasks requiring formal mathematical logic, deep architectural refactoring, or multi-step reasoning. It orchestrates frontier reasoning models including DeepSeek R1, Grok 3, and OpenAI o3 with transparent chain-of-thought derivations.
Auto and explicit model choices
Auto routing directs requests using task classification and real-time provider latency available at that moment. The model selector also exposes individual frontier families, such as OpenAI, Google Gemini, Anthropic Claude, DeepSeek, Grok, Meta Llama, NVIDIA Nemotron, and others.
Image, video, audio, and live experiences
Use dedicated creative studios for specialized media: Image Studio powered by FLUX 1 Dev & FLUX 2 Klein with character consistency; Video Studio powered by Google Veo 3.1 (with 4s, 6s, and 8s durations); Audio Studio powered by Google Lyria 2; and Aura Live for real-time duplex speech.
Privacy and Zero-Data-Retention (ZDR)
All requests transit via TLS 1.3 encrypted tunnels. On eligible commercial and Pro routes, prompts are handled under Zero-Data-Retention agreements ensuring user data is never retained for foundational model retraining.