AIWatch.

Honestly, this is a complete 180-degree turnaround, and I did not see it coming. I would argue that GPT-5 marked something of a low point. Anthropic kept gaining strength, Claude Code soon pulled clearly ahead of OpenAI as a major product, and the GPT-5 launch at the time was received largely negatively. Expectations had been higher, and several mistakes were made along the way. The benchmark fiasco alone is hard to forget.

Anthropic became increasingly powerful in the enterprise and business segments, while OpenAI remained heavily concentrated in the free tier, with 95% of its users there. Meanwhile, the high-revenue business market was increasingly dominated by Anthropic. Broadly speaking, the situation looked like this: OpenAI was strong in consumer, while Anthropic led in business and enterprise, and continued to grow rapidly.

Today, around twelve months later, the picture has completely changed. GPT-5.6 is excellent. Period. Codex is excellent. Period. From a PR perspective, OpenAI is currently doing almost everything right, which is surely also thanks to Tibo and his outstanding community work (just one more reset, Tibo, plz. j/k). OpenAI CFO Sarah Friar has just told employees that annualized revenue in July exceeded total revenue for the entire second quarter.

Codex is becoming increasingly popular among professionals and businesses, which was recently reflected in the explosive growth of its user base. It now has more than ten million users. And in the classic ChatGPT product, OpenAI has reached one billion users.

At the same time, Anthropic, once widely seen as ethically and morally superior, is experiencing a major PR disaster: its positioning against open source, even though its own blog post attempted to soften that stance, a strange shift in sentiment around Opus 5, the PR debacle surrounding Fable 5’s guardrails, the question of whether it should remain in the subscription tier, and much more.

OpenAI is on a winning streak, while Anthropic is currently under enormous pressure. OpenAI has learned from its mistakes, and I have to acknowledge that with genuine respect. The company is now doing so many things well and making so many of the right decisions that I can only take my hat off to them.

Respect, OpenAI. This turnaround deserves to be studied.

Oh, and im super excited what Tibo means by Codex + intelligence too cheap to meter.