Qwen3.8-Max, Inkling-Small, and the Local Inference Shift
About this episode
Alibaba releases 2.4T parameter Qwen3.8-Max at a fraction of rival costs, Thinking Machines Lab's Inkling-Small nearly matches its flagship at one-quarter compute, AMD Ryzen AI Max+ PRO chips bring 70B+ local inference to edge hardware
Chapters
Sources & Citations
Alibaba Qwen Releases Qwen3.8-Max: A 2.4 Trillion Parameter MoE Model | MarkTechPost
www.marktechpost.com
Qwen 3.8 Max Ships: 2.4T MoE, 1M Context, $2/$6 per MTok, Open Weights Next Week | Developers Digest
www.developersdigest.tech
Qwen3.8-Max arrives with a bold claim: it outperforms GPT-5.6 Sol Max and Fable 5 on agentic computer use | VentureBeat
venturebeat.com
Qwen 3.8 Benchmarks: What Alibaba's Table Shows | APIdog
apidog.com
Introducing Inkling-Small — Thinking Machines Lab
thinkingmachines.ai
Inkling-Small Model Card — Thinking Machines Lab
thinkingmachines.ai
Thinking Machines debuts Inkling Small open-source AI model nearing performance of predecessor at about 1/4 size — VentureBeat
venturebeat.com
Thinking Machines Lab Releases Inkling-Small: A 276B Total, 12B Active Open Weights Multimodal MoE Model — MarkTechPost
www.marktechpost.com
X (Twitter)
x.com
AMD Ryzen AI Max+ 395: The Ultimate Local AI Powerhouse Explained
imfounder.com
Trillion-Parameter LLM on an AMD Ryzen™ AI Max+ Cluster
www.amd.com
AI Inference on AMD Ryzen™ AI Max Processor — ROCm Blogs
rocm.blogs.amd.com
Loading transcript…