A Series of Unfortunate Events for OpenAI Users

The piece chronicles a rapid succession of product and pricing moves that have frustrated developers relying on Codex-class models. After Opus 5.5 appeared as a highly efficient competitor, OpenAI shipped GPT-6 Sol - hyped but objectively worse than older releases - followed by a buggier GPT-6.1 “SLO” that matched Opus/Astra on efficiency yet suffered extreme latency (reports of ~2 minutes to first token, 4-minute compactions, and ~20 tokens/sec throughput). Pricing changes compounded the technical problems: a reinstated $200 plan halved its advertised usage multiplier (from 20x to 10x), DevDay replaced clear usage labels with vague “more usage” language, and a $500 tier offered only modestly better allowances. New features like “Dots” were promoted but launched undercooked and irrelevant to many dev workflows.

The argument ties these missteps to competitive pressure from Meta’s Muse (1.8M vs 1.3M iOS downloads in the first 12 days, later surpassing 5M), forcing OpenAI to prioritize consumer growth and ad-driven monetization over paid developer ergonomics. With OpenAI claiming ~1.2 billion weekly users versus roughly 35M Codex/Work weekly reach, consumer retention matters more to valuation, the piece contends, so Codex users are being deprioritized. Prediction: developer subscriptions will degrade further; switching vendors is unattractive because rivals either limit integrations or deliver worse ML work. The conclusion: OpenAI must choose between consumer and enterprise focus or risk a slow decline akin to Perplexity.