The most useful movement in this week’s model rankings happened nowhere near first place. Arena’s preference table did not budge, Anthropic still holds seven of its top eight slots, and OpenRouter’s five busiest models stayed in the same order. The signal is lower down: Gemini 3.6

The most useful thing about this week’s model rankings is that they disagree. Meta’s muse-spark-1.2 (xHigh) has landed at No. 4 on Arena’s text leaderboard, interrupting what would otherwise be an Anthropic sweep of the first eight positions. Meanwhile, OpenRouter’s usage chart barely moved: its

A one-place move near the middle of a usage chart should not decide anyone’s model stack. But Claude Opus 5 passing Step 3.7 Flash on OpenRouter is useful precisely because the margin is small: 1.32 trillion weekly tokens against 1.27 trillion, a gap of roughly 3.

The most important number in this week’s model rankings is not DeepSeek’s first-place badge. It is the 13.61 trillion tokens split between two versions of the same Flash model in OpenRouter’s top three. That is what a production migration looks like when the new release arrives

The most useful thing about this week’s model rankings is also the easiest thing to misread. OpenRouter’s chart is not telling us which model is smartest. It is showing where developers are spending tokens when price, latency, availability, launch promotions, and “good enough” quality all collide. That makes

The most useful LLM leaderboard this week is not a leaderboard at all. It is the gap between two of them. On Arena, where people compare model answers head-to-head, Anthropic holds eight of the top ten positions. On OpenRouter, which measures weekly tokens consumed through its marketplace, Anthropic does not

Model leaderboards invite the wrong kind of certainty. A model moves three places, screenshots circulate, and everyone pretends a single ordinal number has settled an architecture decision. This week’s useful signal came from somewhere less glamorous: the invoice. GPT-5.6 Luna entered OpenRouter’s weekly top 20 at number

The most useful number in this week’s model rankings is not an Elo score. It is 601 billion: the number of tokens developers routed through Claude Opus 5 on OpenRouter in less than a week. That is not a controlled benchmark, and it is not proof that Anthropic has

Leaderboards are most useful when they disagree. This week, OpenRouter says developers poured 612 billion tokens into a model that costs nothing, while Arena says the best answers still come from an Anthropic wall at the top of its preference chart. Those results are not competing verdicts on the same

The most interesting model movement this week is not a new number one. It is a free trial ending without the workload disappearing. Tencent’s Hy3 paid endpoint climbed from sixth to fourth on OpenRouter’s weekly ranking, reaching 3.36 trillion tokens. Its free sibling moved the other way,