Qwen/Qwen3.6-35B-A3B model card

Fetched 2026-10-18. Weights repo created 2026-04-15 (HF API createdAt 2026-04-15T05:59:19Z, commit 995ad96eacd98c81ed38be0c5b274b04031597b0). Card nameplate: first open-weight variant of Qwen3.6, following the February Qwen3.5 series, 35B total and 3B activated, 262,144 context, Apache 2.0, vision-language (text, image, video). The card's benchmark table is a vendor claim: SWE-bench Verified 76.0, SWE-bench Pro 52.1, Terminal-Bench 2.1 54.4, GPQA Diamond 86.7, LiveCodeBench v6 75.4, MMLU-Pro 85.2, AIME 2026 92.7. HF safetensors metadata reports 35,951,822,704 BF16 parameters. The card says the artifacts are compatible with Hugging Face Transformers, vLLM, SGLang and KTransformers.