Web Search

Overview

There are two ways to use web search in staik:

• Server-side tool — add "web_search": true to a chat call; the model calls the tool and staik runs the search and loops the results back. You get a finished answer.
• Direct call — POST /v1/tools/web-search returns raw search results without a model.

Server-side (for agents)

Send "web_search": true (or add {"type": "web_search"} to tools) in a regular /v1/chat/completions call. The model decides when it needs to search, staik runs the search and continues until the model has an answer. Works both with and without stream.

curl https://api.staik.se/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-st-your-key" \
  -d '{
    "model": "qwen3.5:9b",
    "messages": [
      {"role": "user", "content": "What happened recently in EU AI regulation?"}
    ],
    "web_search": true
  }'

If you send your own tools in the same call they are not executed server-side — if the model calls one of your tools, tool_calls are returned to you as usual. Only web_search is run by staik.

Via header (clients without body fields)

Some OpenAI clients and agent tools don't let you add custom fields to the request body. Set the header X-Staik-Web-Search: 1 instead — it enables web_search for the call (equivalent to "web_search": true). Many clients let you set headers per provider or model in their config.

curl https://api.staik.se/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-st-your-key" \
  -H "X-Staik-Web-Search: 1" \
  -d '{
    "model": "qwen3.5:9b",
    "messages": [{"role": "user", "content": "..."}]
  }'

How the loop works

1. The model is given access to the web_search(query, max_results) tool.
2. If it needs current information, it calls the tool.
3. staik runs the search and feeds the results back to the model.
4. The model answers — or searches again (up to an internal limit).

Tokens for all rounds (search calls + final answer) are summed in usage, just like a regular call.

Direct call

If you just want raw search results (no model) — e.g. for your own RAG pipeline — use POST /v1/tools/web-search:

curl https://api.staik.se/v1/tools/web-search \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-st-your-key" \
  -d '{"query": "swedish LLM API", "max_results": 5}'

# => {"results": [{"title": "...", "url": "...", "content": "..."}, ...]}

Privacy

Search runs via a self-hosted search engine on staik's own hardware — no third-party search API key. The query is never logged, the same way we never log prompt text.