Add web search to your WEC Inference calls

+

0/8

🎯 Skill path0/8 earned

AI evals & observability

• 1Prove a model works
• 2Trustworthy JSON
• 3Real test data at scale
• 4Observe & score production
• 5RAG, end to end
• 6Catch regressions in CI
• 7Trace & debug agent tool calls
• 🏆Add live web search

Finish this tutorial, then mark it complete below to earn the skill.

Say you've built a support chatbot or a RAG assistant on WEC. It's solid on your docs and on what the model already knows — but ask it something current ("what's the latest release of X?", today's pricing, a recent change) and it either guesses or tells you its knowledge is out of date. For anything that needs to stay current, that's a real gap.

WEC Inference now has a web-search tool built in. Add it to a normal /v1/chat/completions call and the model can look things up on the live web — the search runs on WiLine's own infrastructure (SearXNG), so nothing goes out to a third-party search vendor. Let's test it.

Reproducibility

Calls hit https://inference.wiline.com/v1/chat/completions with your WEC Inference API key, model Qwen3.5-9B. Nothing to install — the tool is part of the API.

---

Without web search

A normal call, asking something recent:

curl -sS https://inference.wiline.com/v1/chat/completions \
  -H "Authorization: Bearer $WEC_API_KEY" -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen3.5-9B",
    "messages": [{"role": "user", "content": "What is the latest stable version of Docker Engine, and when was it released?"}]
  }'

Figure 1. With no tool, the model won't commit to a version — it says its knowledge is limited to its training cutoff and points you to the release notes instead.

---

With web search

The same request, plus a web_search tool and tool_choice: auto:

curl -sS https://inference.wiline.com/v1/chat/completions \
  -H "Authorization: Bearer $WEC_API_KEY" -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen3.5-9B",
    "messages": [{"role": "user", "content": "What is the latest stable version of Docker Engine, and when was it released?"}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "litellm_web_search",
        "description": "Search the web for current information",
        "parameters": {
          "type": "object",
          "properties": {"query": {"type": "string", "description": "The search query"}},
          "required": ["query"]
        }
      }
    }],
    "tool_choice": "auto"
  }'

Figure 2. Same model, same question, plus the tool — now it returns a concrete version and release date pulled from the live web. The model decided to search, WEC ran the query, and fed the results back before answering.

That's the whole feature: one tool in the request body, current answers out.

---

One thing to watch: tokens

Web search injects the results into your prompt, so a searched call costs more tokens than a bare one (in our test, prompt tokens went from ~26 to ~2,776). Keep tool_choice: auto so the model only searches when it actually needs to — it'll skip search for questions it can already answer.

When to use it

• Yes: current versions, prices, news, anything after the model's training cutoff.
• Skip: stable knowledge the model already has — auto handles that for you.

Point it at a RAG assistant or a chatbot and it can stay current without you wiring up a separate search service.

🎯

Finished this tutorial?

Mark it complete to earn Add live web search on your skill path.