Multi-source search with RRF (Reciprocal Rank Fusion).

Returns ranked results from SearXNG, Tavily, and LinkUp. Use for raw search results without synthesis. No LLM call.

Full research pipeline: search + LLM synthesis with citations.

Pipeline: Multi-source search → Source aggregation → LLM synthesis → Citation formatting

Quick conversational answer using LLM.

No search, direct response from model knowledge. Use for simple factual questions or follow-ups.

Exploratory discovery with knowledge gap analysis.

Identifies what's known and unknown about a topic. Use for cold-start exploration.

Synthesize pre-gathered content into coherent analysis.

Use when you already have sources from other tools.

Deep reasoning with chain-of-thought analysis.

Two modes, picked automatically by whether sources is provided:

• No-sources mode (default): direct chain-of-thought over the model's own knowledge plus optional context. Use for problems the model can reason about without external evidence; depth-controlled via reasoning_depth.

• Sources-aware mode (when sources is non-empty): chain-of-thought synthesis over the pre-gathered sources, with the same shape REST /api/v1/reason produces. reasoning_depth is ignored in this mode — the chain-of-thought prompt is fixed because the reasoning shape is what matters here, not the prose register.

For style variants over pre-gathered sources, call synthesize directly.