Data◆Article · 3:45 · 1m
Qwen 3.8 27B achieves DeepSeek V4-Pro/GPT-5.6 Luna-level performance on the Artificial Analysis Intelligence Index, marking the first time a local model has reached that capability tier.
AINews · Latent Space
Context◆Article
CORS Chat was built to help test Qwen 3.8 27B running in LM Studio on both an M5 MacBook Pro and an NVIDIA DGX Spark.
The author introduces CORS Chat as a tool built the same day, with GPT-5.6-Sol xhigh, specifically to test Qwen 3.8 27B running in LM Studio on two of their machines: an M5 MacBook Pro and an NVIDIA DGX Spark. ✦ AI generated
author · Simon Willison's Weblog · 2026-08-15 · original ↗
I built this today (with GPT-5.6-Sol xhigh) to help test Qwen 3.8 27B running in LM Studio on both my M5 MacBook Pro and an NVIDIA DGX Spark.
Read full article ↗excerpt · fair-use quotation
- ·Built today to test Qwen 3.8 27B
- ·Runs in LM Studio via local models
- ·Created with GPT-5.6-Sol xhigh
- ·Targets two hardware platforms
- ·M5 MacBook Pro primary machine
- ·NVIDIA DGX Spark second platform
- ·Designed to compare cross-hardware
- ·Supports hands-on Qwen 3.8 27B testing
Related moments
FactArticle · 45 words
Qwen3.6, likely the most popular open-weight LLM series using a hybrid design, substitutes Gated DeltaNet layers for Mamba-2 in its non-attention portions.
Sebastian Raschka · Ahead of AI
FactArticle · 48 words
Alibaba's surprise announcement of a 2.4 trillion parameter open-weight Qwen 3.8 model, coming right after Kimi K3's release, shows Chinese labs are leaning further into open-weight releases rather than just maintaining the status quo.
Nathan Lambert · Interconnects