Sounds quite interesting, a big IDE provider optimizing for local AI with their coding harness. Especially that they picked Qwen3.6 over Qwen3.8 because of the thinking needs. Haven't read the full article yet, but sounds really cool.
