Pushing OpenAI Codex to the Limit: 4-Hour PR Grind Burns 9% of Weekly Credits
andimarafioti · x · 2026-08-10
Developer Andrei Marafioti shared an extreme test of OpenAI Codex, where the agent grinded on a single PR for almost 4 hours. This intensive task consumed 9% of his weekly compute credits on the $200/month plan.
He calculated that maxing out the 100% weekly allowance would equate to 15 million output tokens. Based on the rumored pricing for the GPT 5.6 Sol model, this volume of compute would actually cost $450, highlighting the heavy token consumption of autonomous coding agents.
More from coding & agent
- DIRECT: retrieval-based next-question suggestions for conversational agents — burkov · 2026-10-16
- Claude Code and Codex now autocomplete the user's next action — tokenbender · 2026-10-16
- Hermes Agent ships official Reverse Engineer Anything plugin with one-command setup — Teknium · 2026-10-16
- MLExpert: 35B MoE Agent Trained to Autonomously Handle Machine Learning Engineering — Shangheng Du · 2026-10-16
- Cowork-Gym: Alibaba Synthesizes 64.9K Agent Work Environments, Boosting Qwen by 8.7 Points — Anni Zou · 2026-10-16
- MMSearch-OPD: Joint RL and On-Policy Distillation for Long-Horizon Multimodal Search Agents — PekingUniversity · 2026-10-16