Academy
2. The Four Model Families
Identify the four families — frontier, balanced, fast, and specialist — and know which models belong to each.
OUPI organises every available model into four families, so you always know where to look — even when providers release new versions and names change.
• Frontier — The most powerful models, built for hard reasoning, long documents, and complex agent workflows. Examples: Claude Opus 5, GPT-6 Astra, Gemini 3 Pro.
• Balanced — Excellent quality at a significantly lower cost. This is the right default for most professional work: writing, analysis, summaries, everyday code. Examples: Claude Sonnet 5, GPT-5.6 Terra, Mistral Medium, Gemini 3.8 Flash.
• Fast and light — Optimised for short, high-volume, or low-stakes tasks where speed and economy matter more than depth. Examples: Claude Haiku 4.5, GPT-5.6 Luna, Mistral Small, Gemini Flash Lite.
• Specialist — Models with a specific strength: reasoning models that think step-by-step (GPT-5.6 Sol, Magistral, DeepSeek), code models (Codestral, Kimi Code), and live-web models for up-to-date information (Perplexity).
Why four families instead of just picking the "best" model? Because no single model excels at everything, and relying on one provider is a risk. OUPI connects you to Anthropic (Claude), OpenAI (GPT), Google (Gemini), Mistral, DeepSeek, Moonshot (Kimi), xAI (Grok), and Perplexity — all from the same interface. The family structure gives you a stable mental map: even when a model name disappears or a new generation arrives, you can find its replacement by looking at the same family.
Here is a simple decision rule for everyday use:
- Start with a balanced model. It handles the vast majority of professional tasks well and costs a fraction of a frontier model.
- Move up to frontier when the task is genuinely hard — legal reasoning, multi-document synthesis, complex debugging — or when a balanced model has visibly struggled.
- Move down to fast for repetitive, short, or low-stakes work like reformatting, quick translations, or simple Q&A.
- Pick a specialist when the task matches: chain-of-thought calculations (reasoning), heavy code generation (code models), or queries that need live data (web model).
Cost matters: frontier models consume roughly 10 to 30 times more credits per token than fast models. The most economical strategy is to use a balanced model for the bulk of your work and escalate to frontier only for the hard turns. Your balance shows what each exchange consumed, so you can track the difference.
Not sure which family fits? Let OUPI decide. The automatic model selection reads your request and routes it to a suitable model based on the task, length, and your plan. Choose manually only when you have a specific reason — a preferred provider, a very long document, or a task you know is hard.
Open the OUPI chat interface. Start a conversation with a balanced model, ask a moderately complex question, then switch to a frontier model mid-conversation and ask the same question again. Notice how the history is preserved and compare the depth of both answers.
OUPI's models fall into four families: frontier (hardest tasks), balanced (everyday default), fast (high-volume, simple work), and specialist (reasoning, code, live web). Start balanced, escalate to frontier when needed, drop to fast for simple jobs, and pick a specialist when the task calls for it. You can switch mid-conversation without losing history, or let automatic model selection handle the choice for you.