MiMo-V2-Omni is purpose-built for complex multimodal interaction and execution scenarios in the real world. We constructed a fully omnimodal foundation from the ground up — one that natively integrates text, vision, and speech — and deeply couples perception with action through a unified architecture. This not only breaks free from the limitations of conventional models that prioritize understanding over execution, but also equips the model with native capabilities spanning multimodal perception, tool invocation, function execution, and GUI manipulation. With seamless integration into mainstream agent frameworks, MiMo-V2-Omni bridges the gap between comprehension and control, dramatically lowering the barrier to deploying full-stack multimodal agents in production.
返回模型列表
供应商
在多个供应商之间路由请求。复制供应商标识以设置偏好供应商。
供应商缓存命中率上下文
可用率
24 小时AI Gateway 直连请求及各供应商的成功率。
May 31, 18:09
KingsoftCloud100.00%
吞吐量
24 小时AI Gateway 实时流量的 P50 吞吐量，单位为每秒 token 数（TPS）。
No data
延迟
24 小时AI Gateway 实时流量的 P50 首 token 延迟（TTFT），单位为毫秒。
该时段暂无延迟样本。
活动
该模型随时间变化的 token 用量和请求流量。
Token 消耗
May 31
Xiaomi-
KingsoftCloud635
按实际处理请求的供应商拆分 token 用量。区块越大，表示该供应商处理的模型流量越多。
Artificial Analysis
Apps
Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for. View All
Token 排行
No app usage is available for this period.
相关模型
更多模型来自 Xiaomi