Open-Weight LLM Models — Live Ranking, Benchmarks & Specs
Explore, filter and compare open-weight language models with live specs, benchmarks and license data
Qwen-AgentWorld-35B-A3B
Qwen Team (Alibaba Cloud)
Hybrid Gated DeltaNet + Gated Attention MoE - Language World Model (qwen3_5_moe)
Qwen-AgentWorld
Apache License 2.0
Openness Score
1.0
Ornith-1.0-35B
DeepReinforce AI
MoE decoder-only (qwen35moe), post-trained on Qwen 3.5; reasoning model with tool-calling. 35B total / ~3B active (256 experts, 8 routed + 1 shared).
Ornith
MIT License
Openness Score
1.0
Ornith-1.0-9B
DeepReinforce AI
Dense decoder-only (post-trained on Qwen 3.5 9B), reasoning model with tool-calling
Ornith
MIT License
Openness Score
1.0
GLM-5.2
Zhipu AI
MoE with IndexShare sparse attention + MTP layer
GLM
MIT License
Openness Score
1.0
Kimi K2.7 Code
Moonshot AI
Mixture-of-Experts (MoE)
Kimi
Modified MIT License (Kimi)
Openness Score
0.7
Open Weights
Mamba2-Transformer Hybrid Latent Mixture of Experts (LatentMoE) with Multi-Token Prediction (MTP)
Nemotron
OpenMDW License Agreement v1.1
Openness Score
0.7
MiniMax-M3
MiniMaxAI
Native multimodal MoE with MiniMax Sparse Attention (MSA)
MiniMax
MiniMax Community License
Openness Score
0.7
Gemma4-12B
Dense (Unified, encoder-free)
Gemma4
Apache License 2.0
Openness Score
1.0
Qwen3.6-27B
Qwen Team (Alibaba Cloud)
Hybrid Gated DeltaNet + Gated Attention (dense)
Qwen3.6
Apache License 2.0
Openness Score
1.0
Qwen3.6-35B-A3B
Qwen Team (Alibaba Cloud)
Hybrid Gated DeltaNet + Gated Attention MoE
Qwen3.6
Apache License 2.0
Openness Score
1.0
Kimi K2.6
Moonshot AI
Mixture-of-Experts (MoE)
Kimi
Modified MIT License (Kimi)
Openness Score
0.7
Nemotron-3-Nano-4B
NVIDIA
Mamba2-Transformer Hybrid (Nemotron-Hybrid)
Nemotron
NVIDIA Nemotron Open Model License
Openness Score
0.7
Gemma-4-31B
Dense Causal Transformer
Gemma-4
Apache License 2.0
Openness Score
1.0
Gemma4-26B-A4B
Mixture-of-Experts (MoE)
Gemma4
Apache License 2.0
Openness Score
1.0