Clarification: OpenAI's 20% compute claim refers to monitoring overhead, not total capacity
sjgadler · x · 2026-08-20
Context clarifies that the '20% of compute' mentioned in OpenAI's blog refers to monitoring overhead being 20% of the inference compute being monitored, not 20% of OpenAI's total compute. This follows a two-week pause in RL training to harden security and expand monitoring.
Related event: Clarification: OpenAI's 20% Compute Figure Was Misread(3 posts)→
More from Infra
- Data Engineering Warning: Pipeline Success ≠ Data Quality — _jaydeepkarale · 2026-08-20
- Benchmark: DFlash1 vs DFlash2 Speed on Qwen Models — AlpinDale · 2026-08-20
- IJCAI workshop on sustainable AI accepts 9 papers, focuses on efficiency metrics — banazir · 2026-08-20
- SkyPilot Meetup: GB200 Scheduling and Idle GPU Monetization — skypilot_org · 2026-08-20
- ARK research reveals AI inference costs dropped over 99% in 12 months — PeterDiamandis · 2026-08-20
- Qwen3.8-27B Test: Lower KV Cache Quantization Impacts Reasoning Quality — fbms2 · 2026-08-20