Quick answer

For most of the last three years, ChatGPT was untouchable. In April 2026, that story is changing. Claude 4.7 now leads GPT-5 on coding, reasoning, and long-document analysis benchmarks. Most writers and developers who use both say Claude feels noticeably better. OpenAI is still the biggest name — but the "best AI" title is now genuinely contested for the first time.

Something shifted in AI during the first half of 2026. If you read developer forums, you notice a pattern — people who used to default to ChatGPT are switching to Claude, and they are vocal about it. The AI leaderboards have flipped. Here is what actually happened.

What changed in the leaderboard?

The change is not marketing. It is measurable, across the benchmarks that the AI industry has agreed to care about:

  • SWE-Bench Verified (real coding tasks): Claude 4.7 scores 87.4% — GPT-5 scores 79.1%
  • GPQA Diamond (expert-level reasoning): Claude 4.7 scores 93.1% — GPT-5 scores 90.4%
  • Long-document comprehension: Claude 4.7 leads with its 500,000-token context window
  • Blind writing quality tests: users prefer Claude output 61% of the time for long-form prose
  • Instruction following: Claude 4.7 sticks more closely to exact user requests without drift

Why does this matter?

For two years, the narrative was simple — ChatGPT was the best, and everyone else was catching up. That narrative shaped everything: enterprise contracts, developer tooling, integrations, media coverage. When people said "AI", most of them meant ChatGPT. A real leaderboard shift changes procurement decisions, investment flows, and which model becomes the default inside workplace tools.

Is this the end of OpenAI's dominance?

No — but it is the end of its monopoly on "best". OpenAI still has the largest user base by a wide margin, the strongest consumer brand, and the most third-party integrations. ChatGPT is still the AI most people use. But among serious users — developers, researchers, people who build products with AI — the switch to Claude is accelerating. That matters because power users shape which tools become standard over time.

Data point: Anthropic's API revenue roughly tripled between mid-2025 and early 2026. OpenAI is still larger in absolute terms, but Anthropic is now growing faster on the enterprise and developer side — which is where margins are highest.

What do users actually prefer?

In honest user surveys, the picture is nuanced. Casual users tend to prefer ChatGPT because it is faster, has voice mode, and has image generation built in. Professional users — especially developers, writers, and researchers — increasingly prefer Claude because its answers are more careful, its code is more reliable, and its instruction following is more precise. If you only use AI occasionally, the ChatGPT free tier is still excellent. If AI is part of your daily work, Claude 4.7 is worth trying.

Should you switch?

Try both. The easiest test: take three real tasks from your work — a tricky email, a piece of analysis, a coding problem — and run each through ChatGPT and Claude. Notice which one gets you to the finished answer faster, with less back-and-forth. For many people, Claude now wins that test. For others, ChatGPT is still the better fit. The point is that in 2026, you should not assume ChatGPT is best by default. It is a genuine contest.

Related reading

Claude Opus 4.7 — What Is New and Why It Is the Most Capable AI Yet ChatGPT vs Claude vs Gemini — Which AI Is Actually Better? What Is GPT-5 and What Can It Actually Do? OpenAI vs Anthropic — What Is the Difference?

Bottom line

For the first time since ChatGPT launched, there is a real competitor that serious users prefer. That is not hype — it is measurable in benchmarks, revenue, and behaviour. The "best AI" conversation is now genuinely open, and Claude is leading on the metrics that matter to demanding work. Worth paying attention to.