Qwen 3.7 Max costs a fraction of Claude and the gap is closing

When a blog post calls Anthropic's customers simps

The article opens with a lamplighter metaphor and does not tone down from there. The author — an engineer and self-described sci-fi nerd — argues that American AI companies have built a parasocial economy: customers paying premium prices not because frontier models are genuinely superior, but because of brand devotion that resembles the creator-fan dynamic on subscription platforms.

The thesis and the numbers behind it

The core claim is that Chinese models, specifically Qwen 3.7 Max from Alibaba, have closed the practical performance gap with Claude Opus and GPT-5 while costing a fraction of the price. The author's $100 Qwen plan delivers 100,000 credits spanning multiple providers — DeepSeek, Moonshot, MiniMax — compared to what they characterize as rate-limited and overpriced American alternatives.

The author concedes they were once an idolater of OpenAI, then switched — only to commit what they call the cardinal sin of idolatry again with another Western provider before their current Chinese-model conversion. The piece predicts that enterprise customers not constrained by geopolitical regulation will judge that US frontier models have stopped earning their multiplier.

The OnlyFans economy framing: just as creators extract recurring payments through parasocial attachment rather than uniquely irreplaceable content, American AI labs extract $20–$200/month subscriptions from users who could get comparable outputs elsewhere. The investors propping up OpenAI valuations are funding a future that will disappoint them.

What the thread said — four reversals

In the thread, the 193 comments split between people who agreed on model commoditization and people who found the article's style more interesting than its argument.

Reversal 1 — a former BigTech engineer confirms the cheap-model thesis, quietly. The cleanest empirical confirmation came without rhetorical flourish:

"DeepSeek v4 Flash is sufficient for basically all day-to-day coding tasks… my employer hands us all Cursor, I've yet to actually switch it out of 'auto' mode, which mostly runs Composer (their in-house finetune of Kimi 2.5)." — a former BigTech engineer

Reversal 2 — the regulatory moat that pricing cannot cross. The most structurally important counter came concisely:

"Most American companies (regulated ones, definitely) can't dare to touch any Chinese models, though they knew that it makes perfect economic sense. Until the taboo prevails, the cartel get's their flood of profit. That's a cartel protected by regulations." — a commenter on the regulatory constraint

This is simultaneously an endorsement of the author's price-gap claim and an explanation of why it will not collapse the market the way the author predicts. A second commenter noted that Chinese models are already available through AWS Bedrock within US-controlled infrastructure, complicating the cannot-touch-it framing.

Reversal 3 — the OpenAI valuation bear who has been saying this since DeepSeek. A commenter laid out the structural argument for why frontier-lab valuations are untenable with more precision than the article:

"Several companies, most notably OpenAI, are predicated on: 1. There will be an AI moat; and 2. That company will 'win' or 'own' AI. That's the basis of the OpenAI valuation. If that doesn't happen, it's going to be a huge problem to recover sufficient revenue to recoup the investment. And I don't think it will happen." — a commenter on AI commoditization

Reversal 4 — the article itself as evidence against its own argument. Several readers flagged that the prose reads as AI-assisted despite arguing for more authentic AI use, and high-karma commenters found the style grating enough to undermine the substance:

"It's a ludicrous amount of words to say 'I use the cheap models and that makes me very smart.'" — a commenter on the article's style

The real split

The thread's consensus, stripped of the OnlyFans framing, is narrower than the article implies: cheap models are genuinely good enough for many tasks, and commoditization is real. The part the article misses is that the US frontier-lab cartel is partly protected by regulation that a price signal cannot dislodge — and partly by the fact that for the subset of tasks where Opus genuinely outperforms a smaller model, enterprises will keep paying. The question is how large that subset remains as Chinese models scale.