Ask Claude, ChatGPT, Gemini, Grok or Perplexity "which model should I use for this?". Tierpik answers with the model, the exact setting (effort, thinking level or mode), where to find it, which usage limit it draws from, and when to step up, so you stop running out of usage on tasks a lighter model could handle.
Connector URL: https://mcp.tierpik.com/mcp
claude mcp add --transport http tierpik https://mcp.tierpik.com/mcptierpik_recommend — Given a task (free text like 'debug a failing pytest' or 'summarize a 90-page contract') and the user's AI apps/plans, returns the cheapest model + setting that will do the job well in each app, which usage meter it drains, how much allowance it saves vs. the top model, an escalation ladder, and which app to send it to.tierpik_cheat_sheet — Full table of which model and setting to use for ~30 common task types in one AI app on one plan.tierpik_limits — Plans, models, which usage meters exist (5-hour, weekly, per-model, research, images), what drains each, and what to do right now when a limit is hit.tierpik_saving_tips — Ranked habits that cut AI usage (context length, effort levels, batching, agent hygiene), optionally with app-specific steps.tierpik_search — Search Tierpik's guides on AI models, plans and usage limits. Returns guide ids to pass to tierpik_get_guide.tierpik_get_guide — Get the full text of a Tierpik guide by id (from tierpik_search).Read-only. No sign-in. Nothing you send is stored. Model and plan data: Data checked 2 Nov 2026.