OpenAI Speeds Up GPT-6 Astra and Sol by 50% as Subscription Quota Cuts Draw User Backlash

OpenAI raised default output speeds for its GPT-6 Astra and GPT-6.1 Sol models by about 50% on Oct. 6, lifting throughput from 30 to 50 tokens per second for subscribers at no added action. The move follows user complaints after OpenAI halved token quotas on its $200 plan. Research firm SemiAnalysis found Anthropic's daily-driver model quota is worth roughly five times OpenAI's on an API-equivalent basis. For the $200 plans, Anthropic's Claude Fable 5.1 offers about $2,485 in API value versus $2,897 in value exhausted on OpenAI's Astra, while daily models Claude Opus 5.5 and GPT-6.1 Sol diverge more sharply. OpenAI's new $500 plan offers just 21% more Astra capacity than the old $200 tier, and high-tier plans no longer carry a per-dollar token advantage. Existing users retain prior quotas until Oct. 29, when cuts bite. SemiAnalysis estimates Anthropic subscriptions generate about 10% of revenue while consuming over 40% of inference compute.