Compare LLM Models
Select 2-5 models to compare side by side.
Select models to compare
0 / 5 selected
Top downloads
Select 2-5 models
| Attribute |
xiaomi
|
|---|---|
| Company | xiaomi |
| Release Date | 2026-09-22 |
| Parameters (B) | 1020 |
| Architecture | Sparse MoE Transformer |
| Context Window | 1000000 |
| Input Modalities | text, image, audio |
| Open Source | No |
| License | |
| Score | 46 |
| VRAM (GB) | 1280 |
| Compute | 2-node cluster with 16× NVIDIA H100 80GB (tensor parallel 16, expert parallel 16) |
| Benchmark: Agents-Last-Exam | 31.6 |
| Benchmark: AutomationBench | 53.1 |
| Benchmark: CyberGym | 94 |
| Benchmark: DeepSWE | 71.9 |
| Benchmark: GDPval-AA | 1673 |
| Benchmark: Terminal-Bench-2.1 | 89.9 |
| Benchmark: Toolathlon-Verified | 76.9 |
| Pricing (Input) | 0.43 |
| Pricing (Output) | 0.87 |