AI 导读
OpenRouter 为其 Pareto Code 路由推出 Nitro 变体，按实测吞吐量对当前层级内的模型重新排序，把每个请求发送给可用的最快模型。该模式以牺牲部分模型多样性为代价换取明显更高的吞吐量，适用于更看重速度而非价格的使用场景，入口为 openrouter.ai/openrouter/pareto-code:nitro。
正文
Need speed over price? Flip the variant to Nitro. openrouter.ai/openrouter/par…
We re-rank the models in your tier by measured throughput and send each request to the fastest one available, trading some model variety for noticeably higher throughput.
来源：@OpenRouter · x.com