简而言之：在过去一周，开放权重模型在 OpenRouter 上达到了具名模型级 token 总量的 69.1%，而提供商格局则分散在 Anthropic、DeepSeek、Google、OpenAI 以及一个庞大的长尾群体之间。
竞争是一种发现程序。——弗里德里希·哈耶克
而开发者们正在发现开放模型的价值。
OpenRouter 提供了一个观察模型市场的有用视角。¹ 它并非 AI 经济的全部，但它接近 API 前沿——在那里，开发者可以快速切换模型、每日比较性价比，并将每个请求路由到最佳可用选项。
自 2025 年以来，开放模型在 OpenRouter 上的份额急剧增长。在最新的模型层面快照中，开放权重模型在具名的开放与封闭模型 token 总量中占比 69.1%，封闭模型则占 30.9%。
新模型会吸引开发者的关注和大规模测试，随后 token 使用量激增。每次不同模型的新一轮集中发布，都会支撑起一个新的 token 用量平台期。
正如在封闭模型生态中一样，开放模型之间的竞争也意味着快速创新和排行榜的更迭。
DeepSeek 的早期领先地位在 2025 年底和 2026 年初被 MiniMax 和 Kimi 的模型所取代。随后，来自 MiMo、Qwen（阿里巴巴的开放权重模型家族）、Hy3（腾讯的开放权重模型发布）以及 DeepSeek 的发布再次重新洗牌了市场份额。
专注于美国市场的实验室 Arcee 近期表现强劲。
开放模型目前仍只占整体推理的一小部分，但蓬勃的竞争、不断增长的使用量以及实验热潮表明，开发者越来越愿意将生产流量路由到这些模型上。
-
数据来源：OpenRouter 排名和使用数据，基于 OpenRouter 分析数据集中每周 token 用量快照的分析。↩︎
In short : Open-weight models reached 69.1% of named model-level OpenRouter token volume in the latest week, while provider leadership fragmented across Anthropic, DeepSeek, Google, OpenAI, & a large long tail.
Competition is a discovery procedure. — Friedrich Hayek
And developers are discovering the value of open models.
OpenRouter offers a useful view into the model market.1 It is not the whole AI economy. But it is close to the API frontier, where developers can switch models quickly, compare price-performance daily, & route each request to the best available option.
Since 2025, open models have grown sharply on OpenRouter. In the latest model-level snapshot, open-weight models generated 69.1% of named open-versus-closed token volume. Closed models produced 30.9%.
New models attract developer attention & large scale testing, after which token use surges. Each new clustered release of different models sustains a new plateau of token volume.
Just as in the closed-model ecosystem, the competition among open models means rapid innovation & leaderboard changes.
DeepSeek’s early lead gave way to MiniMax & Kimi models in late 2025 & early 2026. Later, launches from MiMo, Qwen, Alibaba’s open-weight model family, Hy3, Tencent’s open-weight model release, & DeepSeek reshuffled share again.
Arcee, a US lab focused, makes a strong appearance recently.
Open models still represent a fraction of overall inference, but the thriving competition, increasing usage, & surge of experimentation suggest developers are increasingly willing to route production traffic to them.
-
Source data: OpenRouter rankings & usage data, analyzed from weekly token-volume snapshots in the OpenRouter analysis dataset. ↩︎