Models from my public MoE training and serving experiments. Notes link to tested setups; support varies by path. Start with the research Space.
-
Qwen/Qwen3-30B-A3B
Text Generation • 31B • Updated • 1.14M • • 983 -
google/gemma-4-26B-A4B
Image-Text-to-Text • 26B • Updated • 76.7k • • 469 -
moonshotai/Kimi-K3
Image-Text-to-Text • 2.8T • Updated • 753k • • 12k -
Qwen/Qwen3-235B-A22B-Instruct-2507
Text Generation • 235B • Updated • 75.8k • • 813