Deep Research API showdown: OpenAI, Perplexity, Parallel, Tavily, Gemini, Valyu compared
bar_raiser333 · reddit · 2026-10-16
A hands-on comparison of six deep research APIs for workflow integration:
- OpenAI Responses API: composable web search, file search, MCP, Code Interpreter; can generate files (retrieve before container expiry); o3/o4-mini-deep-research retired.
- Perplexity Agent API: now recommended over the failure-prone Sonar; depth presets, background tasks, MCP, deliverable sandbox.
- Parallel Task API: custom I/O schemas, structured JSON or Markdown, evidence attached per field, Task Groups for batch runs.
- Tavily: structured output, streaming, citations; includedomains is a preference while excludedomains blocks subdomains; attachments limited to txt/md/json.
- Gemini: plan-approve-execute flow; uses Interactions API with background=true, no structured JSON, can generate charts.
- Valyu: deliverables param outputs XLSX/DOCX/PPTX/PDF/CSV, academic/financial/clinical/patent sources, human-in-the-loop checkpoints, Heavy/Max modes (author reports 2x faster).
Ends with a feature and pricing table — a solid selection guide.
More from coding & agent
- Satire: 19 Claude agents 'reverse-engineer' Blender, forgetting it's already open source — Dan_Jeffries1 · 2026-10-16
- Engineers Should Mine Gong Calls for Ideas That Never Became Tickets — curious_vii · 2026-10-16
- One 80-Word Prompt, Overnight AI Agent Built a Site With 205 Bakeries Mapped — jamesbrooksco · 2026-10-16
- Visual Lean: drag-and-drop GUI for the Natural Numbers Game runs Lean 4 in WASM — MilesCranmer · 2026-10-16
- Fine-Tuned 0.6B Router Beats Its 27B Sibling: Small Specialized Models Win — mshubham · 2026-10-16
- Engineering Leader: AI Got Me Coding Again — I Built Our Feature-Flag System With an Agent — rseroter · 2026-10-16