The claim making rounds today: open-weight models now carry 72% of tokens on OpenRouter. It comes from a single X post with no link attached, and the reason it matters is that OpenRouter's own public rankings page can settle it with one dated capture.
Key facts
- @LeaderX_btc claims 72% of OpenRouter tokens flow to open-weight models — unverified.
- The post also discusses distillation and black-box provider risk — commentary, not data.
- OpenRouter's public rankings page is the primary source that can verify or falsify the share figure.
What's driving the conversation
The post from @LeaderX_btc pairs the 72% figure with a familiar argument: open-weight models are eating routing volume, distillation lets closed labs commoditize their own frontier models, and black-box providers carry concentration risk. Medium velocity. The number is exactly the kind of datapoint the week's larger inference-economics thread from Deedy Das leans on, which is why it is spreading: if open models dominate routed traffic and provider margins are fat, the serving land-grab follows.
Replies are the usual split — people asking for the source, people posting screenshots of OpenRouter's leaderboard that neither confirm nor date the 72% figure.
The substance
What would verify: OpenRouter's public rankings and token-share page, captured with a date, showing open-weight models' aggregate share. The page publishes the underlying data, which makes this one of the rare X claims with a one-click verification path. What stays claim-only until then: the 72% number itself. Do not repeat it as fact — our DeepInfra and router coverage has shown before that adoption narratives outrun the leaderboard.
Why builders are watching
If roughly accurate, the figure says open models are the default on multi-model routers, not the exception — which changes how builders think about portability, pricing pressure, and the alternatives to any single router. It also feeds the privacy conversation: router volume means prompts flow through an intermediary, which is central to what OpenRouter logs and how routers compare on our coverage. Watch for a dated rankings capture this week; either it confirms the share or the claim dies quietly.
Questions
- Do open-weight models really handle 72% of OpenRouter tokens?
- That is a claim from a single X post with no link attached, and DeAI has not verified it. OpenRouter publishes public model rankings and token-share data, so the figure can be confirmed or falsified against a dated capture of that page. Treat 72% as a claim until then.
- Why does OpenRouter token share matter?
- OpenRouter is one of the largest multi-model inference routers, so its token distribution is a proxy for which models builders actually route production traffic to. A high open-weight share means open models are the default on routers rather than the exception.
Sources
About DeAI
DeAI is an independent publication covering open-weight AI models, private inference, and decentralized infrastructure — the tools for running AI you actually control. We test providers on price, privacy, and refusal behavior and publish the numbers, not the vibes. DeAI is powered by Morpheus (mor.org), a decentralized inference marketplace, and covers it on the same terms as every other provider.
Powered by Morpheus and StrandCMS
Morpheus is a decentralized inference marketplace, covered on the same terms as every other provider — we rank it wherever the data lands. StrandCMS is the open-source, agent-first framework this site is built on.
