GPT-6 Astra burns 30 minutes of compute on oversized tasks, then dumps all progress at the limit
CautiousMagazine3591 · reddit · 2026-10-03
A Plus user generating 70+ page documents from 100,000+ characters of source material with GPT-6 Astra found the task runs 30-40 minutes burning heavy compute, then hits the 5-hour usage limit error without outputting a single word. He has to start a new chat with a shrunken task and repeat multiple times.
- Per estimates from ChatGPT and Gemini, each massive run costs roughly $3-$5 in API-equivalent compute; doing this 4 times means OpenAI burned $20 of compute on one $20 Plus subscription project.
- He argues it would be far cheaper for OpenAI to warn "this task will exceed your usage, break it into three parts" instead of letting servers grind for 30 minutes and dumping the progress.
More from Infra
- DGX Spark shortage derails $8,800 donation plan as buyer can't find stock — cyrus_zei · 2026-10-03
- Perplexity to vertically integrate agentic infra on NVIDIA's Vera CPU, ditches x86 — AravSrinivas · 2026-10-03
- 2.3x faster Qwen3.8 27B on RTX 5090: ninfer vs llama.cpp benchmarked across 4 setups — theexile1337 · 2026-10-03
- AirLLM runs a 70B LLM on a single 4GB GPU by streaming layer weights from disk — techNmak · 2026-10-03
- SemiAnalysis asks if GPUs are money printers: TPU v7, Vera Rubin vs Blackwell — AccBalanced · 2026-10-03
- gufo Ships Pre-packaged Windows Build for AMD Strix Halo, 40 tps on Agentic Workloads — hiImMate · 2026-10-03