Leak suggests OpenAI is testing $500/month ChatGPT Pro Max tier
Reports around September 26 indicate OpenAI is preparing a $500-per-month ChatGPT Pro Max tier aimed at high-end agentic workloads, bundling "Fastest Work and Codex."
Around September 26, secondary outlets including bicarait.com and TenBrief cited leaked references suggesting OpenAI is preparing a $500-per-month ChatGPT Pro Max tier for "high-end agentic workloads," described as bundling "Fastest Work and Codex." The package most likely targets long-running, high-throughput, low-latency agent workloads and is reported to use dedicated inference clusters — possibly Cerebras-backed — to remove the noisy-neighbor effect of shared inference environments.
The timing aligns with other observations around the same date: expansion of the Ultrafast serving tier’s documentation, the appearance of an unpublished Standard / Fast / Ultrafast speed selector in the Responses API Playground, and reports of broader rollout around DevDay. Ultrafast was announced August 13 as the first product of OpenAI’s roughly $10 billion three-year compute partnership with Cerebras, remains restricted to GPT-5.6 Sol, and is still waitlist-based. The $500 tier’s formal debut at DevDay has not been confirmed by OpenAI.
Anthropic, by contrast, is leaning on its September 26 announcement of a limited-time free weekly quota reset and Opus 5.5 default promotion to drive Claude paid-conversion, rather than publishing a new $500 tier — both companies appear to be treating "agents plus usage-based pricing" as the next commercial frontier.