Just two days ago, Sam Altman stood before cameras vowing GPT-5 would be the "best model in the world" and hailing it as a "significant step toward AGI." What we got instead was one of the most cringeworthy product launches in recent history, complete with comically flawed charts, a routing system so broken it made the new model seem less intelligent than its predecessor, and user outrage so fierce that OpenAI had to resurrect GPT-4o within a day.

I've been tracking AI progress obsessively since ChatGPT's release, and I've witnessed my fair share of overhyped debuts. But GPT-5 strikes me as different. This isn't just another letdown, it's a glimpse into how desperate OpenAI has become to sustain their exponential progress story when the real gains are incremental at best.

The charts were the most humiliating part. During the livestream, OpenAI displayed graphs so deceptive they instantly spawned a Twitter meme dubbed "chart crime." One showed GPT-5 with a 50% deception rate versus o3's 47.4% (basically the same number) but depicted o3's bar as dramatically taller to portray GPT-5 as superior. Another compared scores of 74.9, 69.1, and 30.8 with bars that had no connection to the actual numbers.

Altman called it a "mega chart screwup" on Twitter, but the damage was done. The irony was lost on no one: the allegedly premier AI company couldn't produce accurate charts for their own event. Ex-OpenAI researcher Nat McAleese, now at Anthropic, smugly remarked that people kept texting him that he "would never have let that plot happen" back when he was their "chart crime police."

But the chart debacle was only the start. GPT-5's headline feature (a "unified" experience with an intelligent router that automatically selects the optimal model) utterly crashed at launch. Reddit was inundated with complaints that GPT-5 seemed "dumber" than GPT-4o. During a damage-control AMA on Friday, Altman finally conceded the router "was out of commission for a chunk of the day" due to what he delicately termed a "sev."

As a result, the supposedly most advanced model ever developed was randomly routing queries to whichever model was still working, making it appear less capable than its predecessor. When your marquee innovation is broken automation that degrades your product, you may be trying too hard.

The routing fiasco reveals something deeper about what GPT-5 really is: a cost-cutting measure disguised as a technological leap. The "unified" model isn't about enhancing user experience. It's about slashing OpenAI's compute bills by routing most queries to cheaper, smaller models while maintaining the illusion users are getting top-tier AI.

This becomes clear when you examine the context window situation. The API version of GPT-5 handles 400,000 tokens, but ChatGPT users get a fraction of that. Free users plummeted from 32,000 tokens to a mere 8,000. Plus users went from 128,000 to 32,000. Meanwhile, Claude provides everyone 200,000 tokens and Gemini offers a million. OpenAI literally degraded the experience for most users while asserting they're delivering the "best model in the world."

User reaction has been scathing. People characterize GPT-5 as feeling "corporate" and "sanitized," more akin to conversing with a customer service bot than an AI. The backlash was so intense that users lobbied during the AMA to permanently resurrect GPT-4o, and Altman had to vow they were "looking into" it. When your product launch necessitates immediately offering access to the prior version, something has gone fundamentally awry.

Throughout this debacle, Altman's hubris has been on full display. His grandiose assertions about GPT-5 being "unimaginable at any previous time in history" ring hollow when the actual product flounders with basic routing decisions. His recent podcast appearances where he discusses feeling "useless" compared to GPT-5 come across as performance art rather than genuine reflection on what the technology truly delivers.

The prediction markets took note. Despite all the hype, Google has now surpassed OpenAI in wagers about which company will have the top model by the end of August. The AI community's faith in OpenAI's trajectory is evidently wavering.

For someone my age starting college as AI supposedly transforms everything, GPT-5's launch is a sobering reminder that most AI progress is incremental, expensive, and frequently overhyped. The companies creating these systems face immense pressure to maintain growth narratives that may not mirror technical reality.

Strip away the marketing jargon and GPT-5's true innovation becomes obvious: it's algorithmic penny-pinching camouflaged as user experience optimization. The router isn't about convenience. It's about cost management. The context window reductions aren't about quality. They're about economics. Even the safety enhancements serve a financial purpose by decreasing human oversight requirements.

This explains why GPT-5's pricing is "aggressively competitive." OpenAI can undercut rivals because users aren't actually running their priciest model for most queries. It's a race to the bottom masquerading as a race to the top.

GPT-5 won't be remembered for its technical feats. It will be remembered as the instant one of AI's most crucial companies prioritized marketing over honesty, cost-cutting over capability, and ego over evidence. In attempting to prove they were still ahead of the curve, OpenAI may have revealed they're actually behind it.

And frankly? Maybe that's precisely what we needed to see.