A multi-hour OpenAI agent re-sends its entire swelling transcript on every turn. Set a compaction threshold and the Responses API summarizes old turns server-side into one encrypted item that carries state forward in far fewer tokens.
Turn On OpenAI's Server-Side Compaction for Hour-Long Agent Runs
🔒 Pro tip · Advanced
Unlock this tip — and 54 more
This is one of 55 advanced, fact-checked tactics reserved for Pro. Get the full 77-tip library, a searchable archive, and a new tip every morning. Free for 7 days, then $9/mo.
Prefer to browse? The 22 Beginner tips are free forever.
More in Context Management
🧠Context Management
Images can be 1,000-2,000+ tokens each; removing stale ones cuts that per turn
Drop the Screenshot Once the Model Has Read It
Images, PDFs, and attachments are charged as tokens and re-sent every turn in a multimodal thread. After the model has described or transcribed one, you usually don't need to keep sending the pixels.
🧠Context Management
50-90% on file-heavy prompts
Paste the Function, Not the Whole File
Most coding questions need 20-40 lines, not your 800-line file. Send the relevant slice plus a one-line note about the rest, and your input shrinks dramatically without hurting the answer.
🧠Context Management
Trims re-sent history; often 20-60% fewer input tokens per turn after a topic switch
Start a New Chat When the Topic Changes
Chat apps re-send your whole conversation with every message. When you switch tasks, the old turns become dead weight you keep paying to re-transmit — even with caching discounts.