7 OpenAI updates this week that cut your API bill, from 12-hour caching to Decisions API
MrCashMahon · reddit · 2026-10-23
A Reddit roundup collects seven OpenAI updates from the week that actually reduce costs:
- 12-hour prompt caching (up from 30 minutes): writing the cache costs 2.5x more upfront, but pays off for calls spaced hours apart.
- Project-level service tier limits: admins can block API projects from using expensive service tiers.
- Decisions API: returns fast, bounded choices without long-form generation, cutting token spend for routing tasks that only need a label or next step.
- Free voice design: prompt-to-voice custom voices are free; you only pay standard audio rates when the model speaks.
- OpenAI models on Google Cloud Marketplace: OpenAI models and Codex are now deployable there.
- GPT-Live-1.1 and Quip voice models: faster responses and better language support.
- GPT-6 Pro: parallel agents on separate task pieces, plus faster outputs.
More from Models
- NVIDIA releases Nemotron 3.5 Bolt E4B, a 4GB open model for on-device conversational agents — NVIDIAAI · 2026-10-23
- NVIDIA releases open Nemotron 3.5 Bolt E4B 4GB with weights, data and recipes — NVIDIAAI · 2026-10-23
- Gemini 'Argon' bug reportedly fails to analyze uploaded audio files — Numerous-Campaign844 · 2026-10-23
- 2-bit quantized DeepSeek 4.1 Flash tested locally: KDL degrades but still works — bakawolf123 · 2026-10-23
- KOL predicts K3.1 benchmarks sight unseen: benchmaxxed, below Sonnet in practice — teortaxesTex · 2026-10-23
- Sam Altman: OpenAI's Best Internal Model Won't Be Released Anytime Soon — rohanpaul_ai · 2026-10-23