OpenAI has started rolling out GPT-6 Astra to ChatGPT users on Pro, Enterprise, and Business Premium plans, with Plus and Business tiers set to follow in the coming days. The model is also live through the OpenAI API, Microsoft Azure, and AWS Bedrock. But the rollout comes with a catch that’s hard to ignore: Astra delivers roughly half the message capacity of its predecessor, GPT-5.6 Sol, across every plan.
On the $200 Pro plan, subscribers get 200 messages per week for GPT-6 Astra Pro, while the $100 Pro plan is limited to 50 per week. Business Premium users also get 50 weekly messages, and Business Standard subscribers are capped at a paltry 15 messages per month. The standard Astra model fares no better. Plus users can expect an estimated 5 to 45 messages per five-hour window, compared to 10 to 100 with Sol. On the Pro 20x plan, that gap widens to 100 to 900 for Astra versus 200 to 2,000 for Sol.
OpenAI attributes the variance to factors like model choice, context length, reasoning depth, tool use, and caching. That’s a lot of variables for users to juggle just to figure out how much they can actually use the thing. Free and Go tier users remain locked out of both frontier models entirely.
The messaging here is interesting. OpenAI spent the last year pushing the idea that each generation would be more capable and more accessible. Astra appears to be more capable, sure — but accessibility is taking a measurable hit. Cutting message quotas in half while keeping prices the same feels like a quiet acknowledgment that serving these larger models is expensive. The company may be betting that enterprise customers will absorb the cost, while individual users learn to ration their prompts. Whether that trade-off holds up once Plus users actually start hitting their limits remains to be seen.
