Model Releases
đ¤ŻQWEN 3.6 35B-A3B IS INSANE AT CODING @unslothai recently dropped Qwen3.6-35B-A3B-GGUF The first open Qwen3.6 model and itâs solid. This MoâŚ
đ¤ŻQWEN 3.6 35B-A3B IS INSANE AT CODING @unslothai recently dropped Qwen3.6-35B-A3B-GGUF The first open Qwen3.6 model and itâs solid. This MoE beast is cooking on benchmarks: đ§ SWE-bench Verified: 73.4 (
đ¤ŻQWEN 3.6 35B-A3B IS INSANE AT CODING @unslothai recently dropped Qwen3.6-35B-A3B-GGUF The first open Qwen3.6 model and itâs solid. This MoE beast is cooking on benchmarks: đ§ SWE-bench Verified: 73.4 (big jump) đ Terminal-Bench 2.0: 51.5 đLiveCodeBench v6: 80.4 â MMLU-Pro: 85.2 + strong GPQA đMultimodal king (MMMU 81.7, VideoMME 86.6) 𤯠Only 3B active parameters â runs smooth đ 262K native context (extendable to 1M) đ¤Elite tool calling & agentic workflows âď¸Unsloth Dynamic 2.0 GGUF quants (2-bit versions hit ~11-13 GB VRAM) đ Perfect for repo-level coding, agents, long-context tasks For frontier-level open-source coding + more.. Try it đđť
Related
- 180 tok/s generation on a 4090 with qwen 3.6. if you're on a 4090 and not running this model yet you're leaving performance on the table. 3BâŚ
- Sorry for the long wait, everyone! As I said, Qwen is going to keep open-sourcing!
- The king reigns supreme This is likely going to be the best finetune of qwen 3.6 35b https://huggingface.co/DJLougen/Ornstein3.6-35B-A3B-GGUâŚ
- I gave two MoE models the same vibe coding challenge Qwen3.6 35B A3B (31.8GB) vs Gemma4 26B A4B (23.3GB) Stack: > Unsloth Q6_K_XL > llama.cpâŚ
Source: Clem Delangue (X) | 2026-04-20