| Add SkillsBench evaluation results | SaylorTwift | open | 0 | disc |
| Listed on OpenModelMap | duola15 | open | 0 | disc |
| Interesting work — mobile AI perspective | 3morixd | open | 0 | disc |
| post training optimization steps | Iamexperimenting | open | 0 | disc |
| Update Terminal Bench 2.0 evaluation notes with methodology details | SaylorTwift | open | 0 | disc |
| Dense 60-120B | rtzurtz | open | 0 | disc |
| delete | rtzurtz | open | 0 | disc |
| Listed on OpenModelMap | duola15 | open | 0 | disc |
| how do people use this thing? | gemstonebro | open | 0 | disc |
| Quantization Aware Training | pcomte | open | 0 | disc |
| Upload WhatsApp Video 2026-05-26 at 15.58.00 (online-audio-converter.com) (1).mp3 | T3R3Z4 | open | 0 | disc |
| Qwen3.7 27B without overthinking | Duonglv | open | 0 | disc |
| We rewrote the multimodal processor for Qwen 3.6 and we reduced latency from 11s to 80ms | zakariaelhjouji | open | 0 | disc |
| Add MMMU-Pro evaluation result | SaylorTwift | open | 0 | disc |
| feat: Add native NVFP4 MTP GGUF community quantization link | nilayparikh | open | 0 | disc |
| set enable thining False but response will output <think> or None .. | titanium2000 | open | 0 | disc |
| 90.0% on SWE-bench Verified using Qwen3.6-27B-FP8 — open-weight, consumer GPUs, full transparency | mrguo | open | 0 | disc |
| When will QWEN 3.6 MAX be released? | ZeroWw | open | 0 | disc |
| Possible future 32 GB single-GPU variant? | vanbukin | open | 0 | disc |
| Qwen3.7-20B-A4B | fenglui | open | 0 | disc |
| When Qwen 3.6 2b / 4b? | sprapp | open | 0 | disc |
| GGUF available — Cerebellum v4 (ablation-guided mixed-precision) | deucebucket | open | 0 | disc |
| function_call do not give me report usually | HeisenbergQ | open | 0 | disc |
| reasoning_effort ? | supersmeh | open | 0 | disc |
| When is 122B-A10B coming? | KeilahElla | open | 0 | disc |
| Fine-tuning Best Practices | study-hjt | open | 0 | disc |
| It seemed to me... | EvilinaMaller | open | 0 | disc |
| Where can I try this? | mindplay | open | 0 | disc |
| Update chat_template.jinja | dluthyc | open | 0 | disc |
| Qwen3.6-27B-AWQ-INT4 for running in vLLM | Duonglv | open | 0 | disc |
| Qwen3.6 seem to fake record too often | anitman | open | 0 | disc |
| MTP Speculative Decoding absolutely no performance gains | Manuun1 | open | 0 | disc |
| Qwen3.6 27b has template issue in ralph loop on opencode: `No user query found in messages` | cropse | open | 0 | disc |
| qwen_6_27b | hjeogndzmvlx | closed | 0 | disc |
| Base model release | aimeri | open | 0 | disc |
| Anyone is having issues with tool calling with the 3.6 family ? (not just 35b or 27b) | LPN64 | open | 0 | disc |
| Qwen3.6-27B is suprisingly good for coding | weisunding | open | 0 | disc |
| what kind of effect if I increase the expert number used in lm studio? | Raffaelelu | open | 0 | disc |
| Anyone noticed the diffrence of sampling parameters between 27B and 35B-A3B (both 3.6) | bash99 | open | 0 | disc |
| Great model, please release ~16B model with FP8 | Duonglv | open | 0 | disc |
| Installation Video and Testing - Step by Step | fahdmirzac | open | 0 | disc |
| MOE 122b and 397b please! | jesleocizi | open | 0 | disc |
| feedback. I just can't hold back my comment! | EvilinaMaller | open | 0 | disc |
| Deploying with sglang, weight name not matching | Kyoma001 | open | 0 | disc |
| 10/10 | qenme | open | 0 | disc |
| Wan when? | kabachuha | open | 0 | disc |
| Add community evaluation results for AIME_2026, GPQA, HLE, HMMT_FEB_2026, MMLU-PRO, SWE-BENCH_PRO, SWE-BENCH_VERIFIED, TERMINAL-BENCH-2.0 | nielsr | open | 0 | disc |
| My deepest respect and admiration goes out to the Qwen team!! | testamentaddress01 | open | 0 | disc |