厂商模型卡;发布时API价格未核实。 · 来源
Estimate from the version's published specs, benchmarks and reviews, 0-5; scores are only comparable within one criteria year. MiMo-V2.6-Pro flagship as launched Sep21UTC/Sep22China, not Flash or UltraSpeed. Launch AA Intelligence Index v4.3.2=46.32; maker model card Terminal-Bench4=34.9% (not Terminal-Bench2.1=89.9%); 1M input; text/image/video/audio inputs. Published launch USD0.435 input/0.87 output per1M givesUSD0.54375 blended(3:1). Speed null: current provider throughput and later test latency cannot establish launch tokens/s. Reviews are disclosed direct tests; DataLLM sells inference, tested viaOpenRouter independently of its gateway; nine simple Python tasks do not establish agentic quality. Later bugs and differing harnesses remain qualified.[6]
MiMo-V2.6 Pro excels at data tasks and fails at longer texts - at least through OpenRouter.
Only one of the three Kubernetes manifests MiMo-V2.6-Pro wrote for us actually served a page.
Nine self-contained Python functions cannot separate a flagship from a competent small model.
| Xiaomi MiMo-V2.6 Pro and Flash Tested: Benchmarks, Pricing · 2026年9月25日 | 0.30 |
|---|---|
| Xiaomi MiMo-V2.6 Review: Benchmarks, Pricing and Real Tests · 2026年9月30日 | -0.40 |
| MiMo v2.6 Review: Three Endpoints, Three 9/9s, and a 16.6x Speed Tax · 2026年10月3日 | 0.30 |
读者上传的照片,与图钉带来源的媒体分开显示。你可以为吃到或看到的内容评分,在餐厅图钉上还能直接在照片上标记菜品。
第一项始终是图钉的来源。总体可信度是各份资料对上方所用开始和结束时间支持程度的加权平均;资料每比最新的一份旧 180 天,权重减半。
小米的公告日期为“2026 年 9 月 22 日”,开篇写道:“今天,我们发布并开源 MiMo-V2.6 系列”;SiliconANGLE[5] 于 9 月 22 日的报道称小米“今天宣布发布并开源”该系列模型。
Xiaomi's[1] model page for Flash: text, image, video and audio input with text output, 1M-token context, 128K max output, and $0.14 per million input tokens (cache miss), $0.0028 on a cache hit and $0.28 per million output tokens.
Lists MiMo-V2.6-Pro (1.02 trillion parameters, 42 billion active) and MiMo-V2.6-Flash (310B, 15 billion active) with a release date of 21 September 2026 - the UTC day the models went live, the evening before Xiaomi's[1] 22 September (China time) announcement.
eWeek adds the third checkpoint, MiMo-V2.6-Distill-Qwen-9B (a fine-tune of Alibaba's Qwen3.5-9B), the 7,000+ RL environments released with the weights, and that Artificial Analysis independently scores Pro at 46 on its Intelligence Index.
SiliconANGLE's same-day report: Xiaomi[1] "today announced the release and open-sourcing of its MiMo-V2.6 series"; gives Pro's agentic scores against Claude Opus 5 (AutomationBench 53.1 v 50.3, Terminal Bench 2.1 89.9 v 89.1) and notes it still trails Claude Fable 5.1 and GPT-6 Astra on aggregate measures.
有遗漏或错误吗?用你自己的话说明:能佐证此图钉的链接、不同的开始或结束日期及理由,或缺失、有误的信息。AI 会对照此图钉的来源进行核实,搜索更好的来源,并添加任何支持你说法的页面。图钉自身的来源仍然最重要。AI 也会查看图片:显示的是别的东西或显示效果差的图片会被移到后面或替换。
原标题: MiMo-V2.6-Pro and MiMo-V2.6-Flash Released