MiMo-V2.6-Pro 1.02T (#1 Open-Source 46.32), MiMo Flash, Qwen3.8-Omni-Max, Kimi K3, Qwen3.6-FP8, Yi-X-Large.
| Rank | Release Date | Model & Lab | Category | Architecture | Score | Action |
|---|---|---|---|---|---|---|
| #51 | Sep 21, 2026 |
MiMo-V2.6-Pro 1.02T Open-Weights Flagship
Xiaomi AI
|
Open-Weights & MoE Engines | Sparse Mixture-of-Experts with Multi-Token Heads | 99.4 | Try → |
| #52 | Sep 22, 2026 |
MiMo-V2.6-Flash UltraSpeed MoE
Xiaomi AI
|
Open-Weights & MoE Engines | Sparse Mixture-of-Experts (Fast Routing) | 98.8 | Try → |
| #53 | Sep 27, 2026 |
Qwen3.8-Omni-Max Multimodal Open MoE
Alibaba Qwen
|
Open-Weights & MoE Engines | Multimodal Sparse Mixture-of-Experts | 98.4 | Try → |
| #54 | Sep 26, 2026 |
Qwen3.8-Omni-Flash Edge Multimodal
Alibaba Qwen
|
Open-Weights & MoE Engines | Compact Multimodal Transformer (Vision & Audio) | 97.6 | Try → |
| #55 | Jul 28, 2026 |
Kimi K3 2.8T Native Context MoE
Moonshot AI
|
Open-Weights & MoE Engines | Sparse Mixture-of-Experts with Multi-Query Attention | 96.8 | Try → |
| #56 | Aug 15, 2026 |
Kimi K2.6 Mathematical Reasoning Engine
Moonshot AI
|
Open-Weights & MoE Engines | Dense CoT Reasoning Transformer | 96.3 | Try → |
| #57 | Aug 31, 2026 |
Qwen3.6-FP8 Native Quantized Enterprise
Alibaba Qwen
|
Open-Weights & MoE Engines | Native FP8 Quantized Transformer | 95.9 | Try → |
| #58 | Aug 24, 2026 |
Yi-X-Large Math & Logic MoE
01.AI
|
Open-Weights & MoE Engines | High-Density Mixture-of-Experts | 95.4 | Try → |
| #59 | Sep 23, 2026 |
Hy-Image-3.5 Ultra-HD Open Diffusion
Tencent AI Lab
|
Open-Weights & MoE Engines | Rectified Flow Multimodal Diffusion Transformer | 95.1 | Try → |
| #60 | Sep 15, 2026 |
DeepSeek-V4 Pro 1.6T Architecture Preview
DeepSeek AI
|
Open-Weights & MoE Engines | 1.6T Multi-Head Latent Attention Sparse MoE | 94.8 | Try → |
| #61 | Sep 18, 2026 |
Llama 3.3 70B Instruct Workhorse
Meta AI
|
Open-Weights & MoE Engines | Dense Auto-regressive Transformer (GQA 8-KV heads) | 94.4 | Try → |
| #62 | Aug 14, 2026 |
Mistral NeMo 2 12B Multilingual
Mistral AI
|
Open-Weights & MoE Engines | Dense Decoder Transformer with TEKKEN Tokenizer | 93.8 | Try → |
| #63 | Sep 10, 2026 |
SmolLM2 1.7B On-Device Reasoning
Hugging Face
|
Open-Weights & MoE Engines | Compact Dense Transformer for Edge Devices | 93.5 | Try → |
| #64 | Sep 8, 2026 |
Gemma 3 27B IT Research Open Weights
Google DeepMind
|
Open-Weights & MoE Engines | Dense Decoder Transformer with Sliding Window Attention | 93.1 | Try → |
| #65 | Sep 7, 2026 |
Gemma 3 9B Flash Edge Engine
Google DeepMind
|
Open-Weights & MoE Engines | Compact Dense Decoder with Knowledge Distillation | 92.6 | Try → |
| #66 | Aug 27, 2026 |
InternLM 3 20B 1M Context Reasoning
Shanghai AI Lab
|
Open-Weights & MoE Engines | Dense Transformer with Long-Context Attention | 92.2 | Try → |
| #67 | Sep 14, 2026 |
Granite 3.1 8B Enterprise Instruct
IBM Research
|
Open-Weights & MoE Engines | Dense Enterprise Decoder Transformer | 91.8 | Try → |
| #68 | Aug 30, 2026 |
Phi-4 Mini 3.8B Synthetic Reasoning
Microsoft Research
|
Open-Weights & MoE Engines | Synthetic-Trained Dense Small Transformer | 91.4 | Try → |