> ## Documentation Index > Fetch the complete documentation index at: https://docs.webrun.ai/llms.txt > Use this file to discover all available pages before exploring further. # Cost Optimization > Reduce costs while maintaining performance WebRun charges for session time and token usage. Three quick changes can cut your costs by 50% or more. ## Quick Wins ### 1. Terminate Sessions Explicitly Set `terminateOnCompletion: true` on your final task, or call terminate explicitly. This alone reduces costs by 50% if you're currently letting sessions timeout. ```javascript theme={null} // Single task - auto-terminates const result = await fetch("https://connect.webrun.ai/start/run-task", { method: "POST", headers: { "Content-Type": "application/json", "Authorization": `Bearer ${API_KEY}` }, body: JSON.stringify({ task: { prompt: "Extract data from this page" } }) }); // Multi-task - terminate on last task const finalTask = await fetch("https://connect.webrun.ai/start/send-message", { method: "POST", headers: { "Content-Type": "application/json", "Authorization": `Bearer ${API_KEY}` }, body: JSON.stringify({ sessionId, message: { actionType: "newTask", prompt: "Final task here", terminateOnCompletion: true } } }); ``` *** ### 2. Write Concise Task Descriptions Shorter descriptions = fewer input tokens. Be specific, not verbose. ❌ **Verbose (150 tokens):** ``` "Please navigate to the Amazon website and then perform a search for wireless keyboards. After you get the results, I need you to look through them carefully and find the ones that have good ratings, preferably above 4 stars, and then return me a list of the top 5 products with their titles, prices, and rating information." ``` ✅ **Concise (35 tokens):** ``` "Go to Amazon, search 'wireless keyboard'. Return top 5 with rating ≥ 4.0 as JSON: title, price, rating." ``` **Impact:** 75% reduction in input tokens for this task. *** ### 3. Set `startingUrl` When Known Skip navigation by providing the starting URL directly. ```javascript theme={null} { "mode": "default", "task": { "prompt": "Search for products", "startingUrl": "https://amazon.com" // Skips "go to amazon.com" step } } ``` **Impact:** Saves 5-10 seconds and \~500-1000 tokens per task. *** ## Additional Optimizations ### Set Appropriate `maxDuration` Limit task duration to prevent runaway costs: ```javascript theme={null} { "task": { "prompt": "Quick search task", "maxDuration": 30000 // 30 seconds max task duration } } ``` **Guidelines:** * Simple searches: 30,000ms (30s) * Form filling: 60,000ms (1 min) * Multi-step workflows: 120,000ms (2 min) *** ### Batch Related Tasks in Single Sessions Reuse sessions for related tasks instead of creating new ones. ❌ **Expensive:** ```javascript theme={null} await runTask("Go to amazon.com"); // New session await runTask("Search for keyboards"); // Another new session await runTask("Get first result"); // Yet another session ``` ✅ **Optimized:** ```javascript theme={null} const session = await startSession("Go to amazon.com"); await sendTask(session, "Search for keyboards"); await sendTask(session, "Get first result", { terminateOnCompletion: true }); ``` **Impact:** 3x reduction in session overhead costs. *** ### Limit Token Usage For simple tasks, reduce token limits: ```javascript theme={null} { "task": { "prompt": "Get product price", "maxInputTokens": 10000, // Lower for simple tasks (default: 100000) "maxOutputTokens": 1000 // Lower for short responses (default: 100000) } } ``` *** ### Use Policies to Prevent Detours Use [automation policies](/concepts/policies) with domain rules to block unnecessary domains and prevent wasted navigation. *** ### Use Custom Proxies for High-Bandwidth Tasks WebRun-managed proxies cost **\$2 per GB** of data transfer. For tasks that download large files or scrape image-heavy pages, use your own proxy to avoid per-GB charges. Only enable proxies when you actually need geographic targeting or IP management — each proxied session adds 2-3 seconds of cold-start latency. ```javascript theme={null} { "proxy": { "source": "custom", "type": "http", "host": "your-proxy.com", "port": "8080" }, "task": { "prompt": "Download all product images from the catalog" } } ``` See the [Proxies guide](/usage-guides/proxies) for configuration details. *** ## Monitoring Costs **Real-time:** Check `usage.cost` in every API response. ```json theme={null} { "success": true, "sessionId": "abc123", "taskId": "xyz789", "type": "task_completed", "data": { "message": "Task completed successfully" }, "usage": { "prompt_tokens": 12450, "completion_tokens": 3200, "total_tokens": 15650, "completion_time": 23.5, "cost": 0.0124 } } ``` **Dashboard:** View historical usage at [app.webrun.ai/usage](https://app.webrun.ai/usage). **During polling:** The `usage` object updates in real-time as tasks execute. *** ## Cost Optimization Checklist * [ ] Use `/start/run-task` for single tasks (auto-terminates) * [ ] Set `terminateOnCompletion: true` on final tasks in multi-task workflows * [ ] Write concise, specific task descriptions * [ ] Provide `startingUrl` when the starting point is known * [ ] Set appropriate `maxDuration` based on task complexity * [ ] Batch related tasks into single sessions * [ ] Limit `maxInputTokens` and `maxOutputTokens` for simple tasks * [ ] Use [policies](/concepts/policies) to prevent accidental navigation * [ ] Use custom proxies instead of WebRun-managed for high-bandwidth tasks * [ ] Monitor `usage.cost` in responses to track spending *** ## Example: Before & After **Before: Expensive** ```javascript theme={null} // Creates 3 separate sessions const result1 = await runTask("Go to amazon.com"); await sleep(5000); // Forgot to terminate, paying for idle time const result2 = await runTask("Please navigate to the search box and then type in 'wireless keyboard' and press enter"); await sleep(5000); const result3 = await runTask("Get me information about the first result"); ``` **After: Optimized** ```javascript theme={null} // Single session, concise tasks, explicit termination const session = await startSession({ mode: "default", task: { prompt: "Go to amazon.com", startingUrl: "https://amazon.com", maxDuration: 60000 // 1 minute max } }); await sendTask(session, "Search 'wireless keyboard'"); await sendTask(session, "Return first result as JSON: title, price, rating", { terminateOnCompletion: true }); ``` **Savings:** \~70% cost reduction from session reuse, concise prompts, and explicit termination. *** Understand WebRun's pricing model Manage session lifecycle Chain tasks efficiently Request minimal, structured responses