Qwen3.6 35B A3B Uncensored

• 🆕 Uncensored, community fine-tune of Alibaba's Qwen3.6 35B A3B MoE
• 🧠 Mixture-of-experts: 35B total parameters, roughly 3B active
• 📏 Base model natively supports up to 262,144-token context
• 👁️ Multimodal model spanning text, images, and video input
• 🔒 Served in a Trusted Execution Environment with hardware attestation
• ⚡ FP8 quantization as served by Venice
• 🔧 Refusal mechanisms stripped from the base model
• 🌐 Web search capability available through Venice

Alibaba

Alibaba Group is a Chinese multinational technology company founded in 1999 and headquartered in Hangzhou, Zhejiang. Originally built around e-commerce and cloud computing, Alibaba has become one of the most prolific contributors to open-weight AI research,…

Qwen3.6 35B A3B Uncensored is a community-modified variant of Alibaba's Qwen3.6-35B-A3B, a mixture-of-experts model with 35B total and roughly 3B active parameters. The "uncensored" descriptor refers to post-training that strips the base model's safety refusal mechanisms; this is not an official Alibaba release but a third-party fine-tune. Venice runs it inside a Trusted Execution Environment with attestation evidence for independent verification, here at FP8 quantization.

Relative to its same-family predecessor Qwen 3.5 35B A3B from February 2026, the Qwen3.6 generation is a multimodal family supporting text, image, and video input, with native context up to 262,144 tokens. The variant inherits these capabilities from the Qwen3.6 base while removing the refusal behavior.

It sits alongside the standard confidential-compute build Qwen 3.6 35B A3B FP8 and the dense Qwen 3.6 27B. Note that uncensored fine-tunes inherit the base model's knowledge rather than adding capabilities; removing refusals does not expand what the model actually knows. Treat outputs accordingly, since safety guardrails have been deliberately removed.