MiMo-V2-Flash is a Mixture-of-Experts (MoE) language model with 309B total parameters and 15B active parameters. Designed for high-speed reasoning and agentic workflows, it utilizes a novel hybrid attention architecture and Multi-Token Prediction (MTP) to achieve state-of-the-art performance while significantly reducing inference costs.
-----
MiMo-V2-Flash creates a new balance between long-context modeling capability and inference efficiency. Key features include:
Hybrid Attention Architecture: Interleaves Sliding Window Attention (SWA) and Global Attention (GA) with a 5:1 ratio and an aggressive 128-token window. This reduces KV-cache storage by nearly 6x while maintaining long-context performance via learnable attention sink bias. Multi-Token Prediction (MTP): Equipped with a lightweight MTP module (0.33B params/block) using dense FFNs. This triples output speed during…
xiaomi/mimo-v2-5-proСемейство: mimoКонтекст (макс.): 1,05 млн токеновВвод: текст; вывод: текстВозможности: рассуждение, инструменты, структурированный выводПоставщики: AI/ML API, Kilo Gateway, NanoGPT, OpenRouter, Standard Compute, Vercel AI Gateway и ещё 1
Источник каталога: https://models.dev/api.json. Получено: 01.11.2026 00:22 UTC. Цены и лимиты — данные источника; перед использованием проверяйте условия поставщика.