这些是读者照片,与图钉带来源的媒体分开显示。
第一项始终是图钉的来源。总体可信度是各份资料对上方所用开始和结束时间支持程度的加权平均;资料每比最新的一份旧 180 天,权重减半。
Inception 的文章《Introducing Mercury 2》写道“Today, we're introducing Mercury 2”,The Decoder[5] 关于发布的报道日期为2026年2月24日;OpenRouter[4] 自2026年3月4日起列出该模型,即该路由器将其上架的日期。
Inception's follow-up announcing Mercury 2 on Azure AI Foundry with the same 1,009 tokens per second, $0.25/$0.75 pricing, 128K context, tool calling and JSON-schema output.[1]
Inception's successor post: Mercury 2.5 adds a 40% intelligence increase over Mercury 2, a 260K context and $0.20 and $0.75 per 1 million tokens, and says thousands of developers and dozens of enterprises had built on Mercury 2.[1]
Router listing: first reasoning diffusion LLM, over 1,000 tokens per second, 128K context, $0.25 and $0.75 per 1 million tokens, released Mar 4, 2026.[1]
Independent launch-day report: 1,009 tokens per second with 1.7 seconds end-to-end latency against 14.4 seconds for Gemini 3 Flash and 23.4 seconds for Claude Haiku 4.5 with reasoning, pricing compared with both, and the $50 million raise last November.[1]
有遗漏或错误吗?用你自己的话说明:能佐证此图钉的链接、不同的开始或结束日期及理由,或缺失、有误的信息。AI 会对照此图钉的来源进行核实,搜索更好的来源,并添加任何支持你说法的页面。图钉自身的来源仍然最重要。AI 也会查看图片:显示的是别的东西或显示效果差的图片会被移到后面或替换。
原标题: Mercury 2 Released