The Qwen3.8 quantizations by Mirai Labs.
AI & ML interests
On-device AI
Recent Activity
Organization Card
Welcome to Mirai Labs!
Here we release optimized language models, quantized model variants, inference tools, benchmarks, and research projects focused on high-performance local AI for Apple-Silicon optimized inference. Feel free to visit Our Website to learn more and try our latest models!
models 33
trymirai/d3-flash-M
Zero-Shot Classification • 4B • Updated • 146
trymirai/Qwen3.8-27B-XS
Text Generation • 7B • Updated • 33 • 2
trymirai/Qwen3.8-27B-S-6GiB
Text Generation • 6B • Updated • 19
trymirai/Qwen3.8-27B-S
Text Generation • 9B • Updated • 670 • 50
trymirai/Qwen3.8-27B-M
Text Generation • 14B • Updated • 2.15k • 10
trymirai/Qwen3.8-27B-L
Text Generation • 27B • Updated • 1.19k • 3
trymirai/LFM2.5-2.6B-M
Text Generation • 1B • Updated • 452
trymirai/LFM2.5-2.6B-L
Text Generation • 3B • Updated • 477
trymirai/LFM2.5-1.2B-Thinking-L
Text Generation • 1B • Updated • 353
trymirai/LFM2.5-1.2B-Thinking-M
Text Generation • 0.6B • Updated • 393 • 1