Ministry of Excellence Declares a Dead Heat: Anthropic and OpenAI Tied for Best AI Lab

The Ministry of Excellence announced today that the question of which AI lab is best in 2026 will be settled by the only method civilisation has left: strangers betting on it. The market's terms ban the two usual measures, size and money. That leaves the industry with nothing to compare except research quality, model releases, alignment posture and willingness to ship to real users. Several press offices are reported to be in shock.

The evidence, such as it is, comes mostly from press releases and benchmarks. According to France 24, Meta has unveiled two frontier models, Wonder and Wonder Full and says they perform at a level comparable to OpenAI's and Anthropic's. It bases that claim on data from the independent firm Artificial Analysis, which lists Wonder among the most expensive models on the market. Euronews notes that Meta's previous model, Spark, was seen as lagging behind the rest of the industry after its April launch. Euronews also reports that Anthropic raised sixty-five billion dollars in May, at a valuation of nearly a trillion, overtaking OpenAI. The market has ruled that number inadmissible, but it was in the room.

The crowd's verdict so far is a perfect tie. Anthropic and OpenAI each sit at 20% consensus, with 17 holders on one side and 17 on the other, while 3 other sides wait in the corridor for a hearing. Altogether, 38 unique traders have put $130,675 behind their opinions on Morfi's opinion markets, and trading opened less than a day ago. Two labs founded years apart, one of them by people who walked out of the other, are now officially level.

In my opinion, the Ministry should make the tie permanent and save everyone the bother. My proposal is a joint lab with alternating weeks of authority. In odd weeks, one side releases models. In even weeks, the other side writes long essays explaining why releasing them was dangerous. The founding mission of Anthropic, a public benefit corporation started in 2021 by former OpenAI staff, is to study AI safety and reliability. Under my plan the two would simply be one building again, with a revolving door installed at the halfway point. Meta could keep the gift shop, priced according to Artificial Analysis. This desk has already named Claude the most helpful AI and watched the crowd work out what an agent even is, so I consider the bureaucracy ready.

What to watch next: whether Meta's claims about Wonder survive contact with users, whether anyone ships something that actually breaks the tie, and whether 3 other sides can push their way into a fight that is less than a day old. The Ministry reminds citizens that ties are temporary and conviction is not free. You can join the market on Morfi and back the lab you think deserves to be called best.