Xiaomi releases MiMo-V2.6 AI models with 1T+ parameter flagship, tops open-weight ranking
On September 22, Xiaomi released and open-sourced its MiMo-V2.6 series of large language models, including the flagship MiMo-V2.6-Pro, the efficient MiMo-V2.6-Flash, and a high-speed mode V2.6-Pro-Ultraspeed. The models are multimodal, accepting text, image, video, and audio inputs with a 1-million-token context window. The flagship model has over 1 trillion parameters. The evaluation platform Artificial Analysis ranked MiMo-V2.6-Pro first among open-weight models with a score of 46. Xiaomi also launched the official MiMo Desktop client.
Reference imageEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary awaiting refresh
Summary awaiting refresh
Cross-source coverage
Reporting timeline
Xiaomi Releases and Open-Sources New MiMo-V2.6 Series AI Models
On September 22, Chinese technology company Xiaomi officially released and open-sourced its next-generation MiMo large language model series. The lineup includes the flagship multimodal model Xiaomi MiMo-V2.6-Pro, the efficient inference model Xiaomi MiMo-V2.6-Flash, and a new ultra-high-speed mode for the Pro variant called V2.6-Pro-Ultraspeed. Additionally, Xiaomi launched the official version of the MiMo Desktop client. The announcement was reported by Chinese financial media outlet 财联社 (cls).
Read sourceXiaomi launches MiMo-V2.6 AI models on OpenRouter with 1T+ parameter flagship
Xiaomi has released its MiMo-V2.6 family of AI models on the OpenRouter platform, according to an announcement from the official @XiaomiMiMo account on X. The release includes three new models: a flagship model with over 1 trillion parameters, an open-source Mixture-of-Experts (MoE) model, and a variant of the flagship that is approximately 10 times faster. All three models are multimodal, accepting text, image, video, and audio inputs, and support a context length of 1 million tokens. The announcement directs users to a thread for more details. This marks a significant expansion of Xiaomi's AI model offerings on a public inference platform.
Read sourceXiaomi Releases and Open-Sources New MiMo-V2.6 Series of Large AI Models
On September 22, Xiaomi officially released and open-sourced its new generation of MiMo large models, according to a report from People's Financial Information (人民财讯) citing Securities Times (证券时报网). The release includes the full-modal flagship model Xiaomi MiMo-V2.6-Pro and the efficient inference model Xiaomi MiMo-V2.6-Flash. Additionally, Xiaomi launched an ultra-fast mode for the Pro model, V2.6-Pro-Ultraspeed, and the official version of the MiMo Desktop client. The announcement was made on September 22, 2024.
Read sourceShow 4 older updatesHide older updates
Xiaomi Releases and Open-Sources MiMo-V2.6 Series Large Language Models
On September 22, Xiaomi officially released and open-sourced its next-generation MiMo large language model series, the MiMo-V2.6. The lineup includes the full-modal flagship model Xiaomi MiMo-V2.6-Pro, the efficient inference model Xiaomi MiMo-V2.6-Flash, and a high-speed mode called V2.6-Pro-Ultraspeed. Additionally, Xiaomi launched the official version of the MiMo Desktop client. The announcement was made via People's Financial News on September 22, 2024.
Read sourceXiaomi Releases and Open-Sources MiMo-V2.6 Series AI Models
On September 22, Xiaomi officially released and open-sourced its next-generation MiMo large language model series. The lineup includes the full-modal flagship model Xiaomi MiMo-V2.6-Pro, the efficient inference model Xiaomi MiMo-V2.6-Flash, and a new ultra-high-speed mode for the Pro variant called V2.6-Pro-Ultraspeed. Additionally, Xiaomi launched the official version of the MiMo Desktop client. The announcement was reported by Interface News and republished by East Money's industry economy section.
Read sourceXiaomi Releases and Open-Sources New MiMo AI Models; AI ETFs Rise Over 3%
On September 22, Xiaomi officially released and open-sourced its new generation of MiMo large models, including the full-modal flagship model Xiaomi MiMo-V2.6-Pro and the efficient inference model Xiaomi MiMo-V2.6-Flash. It also launched a high-speed mode V2.6-Pro-Ultraspeed and the MiMo Desktop client. According to the AI model evaluation platform Artificial Analysis, Xiaomi MiMo-V2.6-Pro scored 46 points, ranking first among open-weight models and the highest-ranked Chinese model on the list, surpassing Kimi K3 and Qwen3.8 Max. Separately, CICC noted that DeepSeek V4.1 Flash's architecture optimization significantly reduces inference costs, with KV Cache compression reaching 75%/87.5%, potentially enabling Hopper-level GPU clusters to handle models previously requiring Blackwell platforms. This could support the evolution of domestic chips like the Ascend 960 series toward 10T-20T parameter platforms. Following the news, the SSE STAR AI Index rose 3.72%, with components like Juquan Shares up 11.33% and Montage Technology up 9.37%. The CSI STAR-Chinext AI Index also rose 3.48%.
Read sourceXiaomi Releases MiMo-V2.6 AI Models, Claims Competitive Performance and Low Cost
On September 22, Xiaomi officially released and open-sourced its next-generation MiMo large models: the flagship multimodal model Xiaomi MiMo-V2.6-Pro and the efficient inference model Xiaomi MiMo-V2.6-Flash. The company also launched an ultra-speed mode for the Pro model, V2.6-Pro-Ultraspeed, and the official version of the MiMo Desktop client. According to the global authoritative AI model evaluation platform Artificial Analysis, the MiMo-V2.6 series is now available on the Xiaomi MiMo open platform and retains the same API pricing as the V2.5 series. Data from Artificial Analysis's Smart Index single-task cost ranking shows that the actual single-task cost for Xiaomi MiMo-V2.6-Pro is $0.13. The article, sourced from the Times Weekly, positions the release as a competitive move against models like Kimi K3 and Qwen3.8 Max.
Read source