4 ms·
Xiaomi releases MiMo-v2.5 Family weights with strong coding and agent benchmarks
- steveharing1 6mo agoIt was released around a week ago but Xiaomi Open Sourced it Now & its MIT Licensed and available om HuggingFace
- landtuna 6mo agoUgh, the slop. "That's not a benchmark row. That's a story."
- Alifatisk 6mo agoI’ve been curious about MiMo models, they score so high on wide range of benchmarks, yet, not much is talked about it. At least in my bubble. Is it a serious contender? What is the model for?
- ilqr_jb 6mo ago[dead]
- swiftcoder 6mo agoHas anyone run the numbers on Xiaomi's token subscription prices? The subscriptions are all priced in terms of "credits", and its not clear how credits map to tokens
- esperent 6mo agoI thought I had a high tolerance for slop but this is literally painful to read. Here's a slightly better source that looks like a human at least reviewed it before publishing. https://venturebeat.com/ai/open-source-xiaomi-mimo-v2-5-and-v2-5-pro-are-among-the-most-efficient-and-affordable-at-agentic-claw-tasks https://venturebeat.com/ai/open-source-xiaomi-mimo-v2-5-and-...
- spwa4 6mo agoTLDR: 2 models, 1 trillion parameters total, 42M active, 1M (MiMo-V2.5-Pro) or 256K (MiMo-V2.5-Pro-Base) context length, available on huggingface here: https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro
- steveharing1 6mo agoLets see when the community come up with more quantized versions. Waiting for Unsloth's version