Chinese content小米发布 MiMo-V2.6 Pro 与 Flash，开源权重对标 GPT-5.6
小米把 MiMo-V2.6 整套开源了，Pro 在智能体基准上打平 Claude Opus 5，本地跑前沿模型的门槛又降了。
Latest
Every published item, in order · 78 items
Sep 23
Xiaomi MiMo-V3 to Adopt New Architecture with HySparse 2
The architecture targets agentic workloads, where each turn appends long observations and prefill cost, cache size and retrieval accuracy sit on the critical path simultaneously. Switching from block-level to token-level selection and folding SWA into a forced recent window gives local and global tokens one shared KV cache. Developers building long-context agents should watch MiMo-V3's release, especially against open-source alternatives like DeepSeek and Qwen in the same weight class.
Chinese contentClaude Opus 5.5、GPT-6 Sol/Luna 同日发布，Mimo v2.6、Grok 4.7 齐上阵
一天之内 Opus 5.5、GPT-6 Sol/Luna、Grok 4.7、Mimo v2.6 全来了，Opus 5.5 还降了 40% 成本，这篇帮你梳理各家提升方向和 benchmark。
Xiaomi releases MiMo-V2.6 multimodal model series
The open publication of real-time training dashboards and technical reports is what sets this release apart, setting a transparency bar few open-source model teams currently meet. Developers tracking open-weight multimodal models should benchmark MiMo-V2.6 against Qwen and DeepSeek releases, as Xiaomi's hardware-software integration could shape public-environment deployments across Asia.
Chinese content开源 MiMo v2.6 成为性价比新标杆，代理成本仅为 Grok 4.7 的 1/57
小米的 MiMo v2.6 Flash 跑分接近 Grok 4.7，成本却只有 1/57，做 agent 的可以看看这份 60 任务实测。
Sep 22
Xiaomi MiMo-V2.6-Pro debuts as top open weights model on Artificial Analysis index
At US$0.435 per 1M input tokens, MiMo-V2.6-Pro undercuts most frontier closed models, and open weights let teams in cost-sensitive Southeast Asian markets self-host or deploy via API.
Chinese content小米 MiMo-V2.6-Pro 登顶开源权重模型，1T 参数训练仅花 300 万美元
小米 MiMo-V2.6-Pro 把开源权重第一的位置抢下来了，1T 参数训练只花 300 万美元，做本地部署的可以看看这套权重。
Chinese content小米 MiMo-v2.6-flash 实测：比 DeepSeek 闲时价更低，输出更快
有人实测 MiMo-v2.6-flash 比 DeepSeek 闲时还便宜、速度还快，预算敏感的话可以去比比价。
Chinese contentXiaomi 发布 MiMo-V2.6，将开源 RL 训练全程
Xiaomi 把 RL 训练全开源，Pro 版跑赢 Qwen3.8 Max，价格比涨价后的 DeepSeek 更低。
Chinese content小米开源多模态模型 MiMo-V2.6-Pro，智能指数 46 居开源第一
小米开源了 MiMo-V2.6-Pro，得分追平 GPT-5.6-Sol Max，价格只有零头，权重可直接下载。
Chinese content小米开源全模态模型 MiMo-V2.6：AA 指数 46 分超 Kimi K3
小米把 MiMo-V2.6 权重全开源了，Pro 在 AA 指数拿 46 分压过 Kimi K3，价格只有海外模型的 1/20。
IT之家Source
Chinese content小米 MiMo-V2.6 公布：或为开源团队迄今最大单次 RL 训练
小米把 MiMo-V2.6 的 RL 训练细节全公开，7K 个环境和训练框架一并开源，做 Agent RL 的直接去读技术报告。
Xiaomi releases MiMo-V2.6-Pro model
Available on OpenRouter, these models offer alternatives to agentic coding solutions from US companies like OpenAI and Anthropic.
Xiaomi's MiMo-V2.6 RL training
MiMo-V2.6 demonstrates massive RL scaling with 2B tokens per step, offering insights for developers working on large-scale AI training.
Xiaomi MiMo-V2.6-Pro debuts in Code Arena rankings
Xiaomi's MiMo-V2.6-Pro offers competitive performance at #3 among open-source models, making it a cost-effective alternative to proprietary AI coding assistants.
Xiaomi releases MiMo-V2.6-Pro model
MiMo-V2.6-Pro offers top-tier performance among open weights models at competitive pricing, making it attractive for developers in cost-sensitive markets.
Sep 18
Sep 17
Chinese content小米发布MiMo-V2.6，研究多任务代理RL的扩展性
小米搞了个叫MiMo-V2.6的模型，在研究多任务代理RL怎么扩展，计算量很大，每步20亿令牌，用了很多提示和滚动，还混合了多种环境，挺有意思的。
Xiaomi MiMo-V2.6 model training
For users interested in large language models, Xiaomi's MiMo-V2.6 is a significant update that demonstrates advancements in reinforcement learning and multi-task capabilities.
IT之家Source
Xiaomi trains MiMo-2.6 model
For users looking for affordable AI models, mimo-v2.5's strong performance on OpenRouter and mimo-v2.6's improved results show Xiaomi's models remain competitive in price and capability compared to alternatives like GLM-5.3-Flash.
Xiaomi MiMo team streams RL training of new model MiMo-V2.6
For users interested in the technical details of AI model training, this provides a rare look at the real-time costs and processes involved in scaling RL.
Chinese content小米 MiMo-V2.6 模型 RL 训练细节公开
小米开源了他们新模型 MiMo-V2.6 的训练细节，能让你了解如何用 RL 构建开源 AI，和之前的方法比，他们把计算、环境和评估都做了更大规模的扩展。
Sep 2
AI models drive market growth
These models represent key developments in AI performance that developers and businesses in ASEAN should consider for implementation.
Aug 27
JIT-Agent: Harness Intelligence Model
JIT-Agent improves performance across multiple model families including DeepSeek V4 and Qwen3.6, offering a scalable alternative to manual harness design.
Jul 15
Chinese content小米 MiMo 通过国家大模型备案，完成全系列模型矩阵闭环
小米的 MiMo 模型正式备案了，端侧大模型覆盖语言、多模态、语音多种能力，还计划砸 600 亿搞 AI，快来看看他们的布局。
IT之家Source
Jul 9
Chinese contentArena周榜：claude-fable-5蝉联榜首，国产模型稳定中上游
这期榜单告诉你哪家模型综合最强、代码最好、数学最牛。国产的qwen3.7-max-preview在代码榜进了前十，值得看看。
Jul 7
Chinese content小米发布 MiMo-V2.5-ASR 语音模型，定价 0.5 元/小时
小米新出的语音模型 MiMo-V2.5-ASR 能听方言、混中英文、扛噪音，每小时才五毛钱，适合做语音转录的朋友试试。
IT之家Source
Jun 18
Models17:43
Chinese content小米开源全屋智能 AI 方案 Xiaomi Miloco 2.0，基于 MiMo 大模型实现主动智能小米把全屋智能 Agent 开源了，基于 MiMo 大模型，能认人、记习惯、主动干活，自己搭一套试试。
Chinese content用户实测：glm 5.2 在 COLA 上优于 deepseek 和 mimo，稳定性和速度待改进
如果你在找国产模型，glm 5.2 在 COLA 上比 deepseek 和 mimo 强，但别着急用，现在慢还不稳定。
Jun 10
Chinese contentMiMo V2.5 Pro UltraSpeed 实测：每秒超1000 Token
MiMo 把万亿参数模型的推理速度拉到 1000+ Token/s，做 Agent 和实时交互的开发者值得关注——速度提升直接缩短任务等待时间，并发场景下体验翻倍。
Jun 9
Models20:09
Chinese content小米 MiMo-V2.5-Pro-UltraSpeed 1T 参数模型生成速度达 1000 TPS1T 参数模型跑出 1000 TPS 的生成速度，对追求低延迟推理的开发者来说是个值得关注的指标，建议申请试用实测质量。
Chinese contentMiMo V2.5 Pro UltraSpeed 每秒1000 Token，万亿参数模型实测
万亿参数模型跑出 1000+ Token/s 的速度，做 Agent 和实时交互的开发者可以直接体验，效率提升肉眼可见。
Chinese content小米 MiMo V2.5-Pro-UltraSpeed 突破万亿参数模型千 tokens/s 输出
万亿参数模型首次达到千 tokens/s 输出，做大规模推理或实时 AI 应用的团队可以直接申请体验，看看能否真正落地到生产环境。
IT之家Source
May 17
Chinese content开源模型爆发：Gemma 4、DeepSeek V4、Kimi K2.6 等密集发布
开源模型一个月内连发五款旗舰，做模型选型或研究的团队可以直接参考 CAISI 的 V4 评估对比，省去自己跑 benchmark 的时间。
May 14
Chinese content小米MiMo-V2.5-Pro登顶多项开源模型排行榜
小米MiMo-V2.5-Pro在多个高难度榜单中超越众多闭源模型，做模型选型或关注开源生态的开发者值得关注——它证明了开源模型在核心智能和实际编码任务上已能媲美顶级闭源方案。
