MiMo-V2-Omni-0327
Xiaomi
描述
MiMo-V2-Omni is Xiaomi's omni foundation model uniting frontier multimodal understanding with strong agentic capability. It fuses dedicated image, video, and audio encoders into a single shared backbone, processing all modalities simultaneously. Natively supports structured tool calling, function execution, and UI grounding. Supports over 10 hours of continuous audio understanding and 256K token context window.
发布日期
2026-03-27
参数规模
—
上下文长度
262K
支持模态
audio, image, pdf, text, video
能力雷达图
33
general
40
coding
86
reasoning
56
science估算
100
agents
85
multimodal
Science 在缺少专门科学评测时使用推理能力代理估算。
排行榜排名
基准测试分数 (LLM Stats)
Agents
GDPval-AA
1410.00 / 3000自报
PinchBench
81.2%自报
Claw-Eval
54.8%自报
MM-BrowserComp
52.0%自报
OmniGAIA
49.8%自报
Code
SWE-Bench Verified
74.8%自报
AA 评测指数
Intelligence Index36.4
Tau20.9
Gpqa0.9
Ifbench0.7
Lcr0.6
Scicode0.4
Terminalbench Hard0.4
Hle0.2
LLM Stats 分类评分
Legal100
Finance100
General100
Reasoning100
Agents100
Frontend Development70
Code70
Coding70
定价
输入价格$0.4 / 1M tokens
输出价格$2 / 1M tokens
混合价格(3:1)$0.8 / 1M tokens
缓存读取价格$0.08 / 1M tokens
速度
Tokens/秒80.3
首Token延迟2.22s
首回答延迟27.13s
供应商价格排行
供应商价格排行
1 个供应商
供应商输入输出
1Xiaomi主要
$0.4
$2
比较该模型在不同 API 供应商之间的定价。