AI reasoning modes

Same chat. Four ways to think.

Pick the mode that matches the question. Three seconds for a daily check-in. Twenty-five for a multi-year arc. Thirty for a council of seven specialists. Or audit somebody else’s reading, claim by claim.

Free tier

Normal mode

Day-to-day questions. Cited. ~3 seconds.

Single-topic questions, daily readings, follow-ups, basic interpretive work.

Model: Claude Sonnet 4.6
Latency: ~3.5 seconds

See Normal mode

Pro tier

Deep Research mode

Long-arc questions. Multi-pass reasoning. ~25 seconds, structured brief out.

Career trajectory, multi-year dasha analysis, life-arc questions, anything that warrants a brief instead of a chat reply.

Model: Claude Opus 4.7
Latency: ~25 seconds

See Deep Research mode

Max tier

Max Council mode

Seven specialists agree. Or they don’t — and you see why.

Big life decisions — marriage, surgery, business launch, major relocation. Where you want every classical school weighing in.

Model: 7 × Sonnet 4.6 → Opus 4.7
Latency: ~25-35 seconds

See Max mode

Max tier

Check Your Pandit mode

Paste any reading. We grade every claim. Green, amber, red.

You got a reading elsewhere and want a sober second opinion. Or you are comparing two pandits before booking.

Model: Claude Opus 4.7
Latency: ~20 seconds

See Check Pandit mode

You don’t pick the model. You pick the mode.

Behind every mode, Haiku 4.5 first triages your question in under 500ms and routes it to the right specialist agent. The mode you pick decides the depth (one reply vs a brief vs a council verdict), the latency budget, and the tier required. The right model for each step is chosen for you.

How our AI works

Try them on your own chart

Normal mode is free. No signup needed for your Kundli.

Generate my Kundli