Mesh routing build: Qwen3-30B-A3B MoE weights configured (not trained) by THOX with the MeshStack expansion profile and Knowledge Pack v0.1. 53.75 tok/s, 14.99 GB loaded, grounding 8/8. Apache-2.0.
MeshGroup workgroup build: Qwen3-30B-A3B weights configured (not trained) by THOX with the workgroup profile, prompt pack, and Knowledge Pack v0.1. 53.68 tok/s, grounding 8/8. Apache-2.0.
Mesh workstation model: Qwen3-30B-A3B MoE weights configured (not trained) by THOX with the Mesh system profile and Knowledge Pack v0.1. 11.14 tok/s CPU, 19.56 GB, grounding 8/8. Apache-2.0.
Mesh agent build: Qwen3-30B-A3B weights configured (not trained) by THOX with the agent system profile, reasoning prompt pack, and Knowledge Pack v0.1. 52.23 tok/s, grounding 8/8. Apache-2.0.
QwenLong-L1.5, a long-context reasoning model built upon Qwen3-30B-A3B-Thinking, augmented with memory mechanisms to process tasks far beyond its physical context window.
Qwen3-30B-A3B-Instruct-2507 has the following features: - Type: Causal Language Models - Training Stage: Pretraining & Post-training - Number of Parameters: 30.5B in total and 3.3B activated - Number of Paramaters (Non-Embedding): 29.9B - Number of Layers
This model is a distilled version of Qwen/Qwen3-30B-A3B-Instruct designed to inherit the reasoning and behavioral characteristics of its much larger teacher model, deepseek-ai/DeepSeek-V3.1.
This repo contains both the Q4_K_XL version of Qwen3-30B-A3B-Instruct-2507 and Qwen3-30B-A3B-Thinking-2507
包含3个量化版本GGUF：Qwen3-30B-A3B-Q8_0，Qwen3-30B-A3B-Q5_K_M，Qwen3-30B-A3B-F16
Unsloth Dynamic 2.0 Quants achieves superior accuracy & SOTA quantization performance. UD-Q4_K_XL provided.
https://huggingface.co/burtenshaw/Qwen3-30B-A3B-python-coder
Qwen3-16B-A3B is a rendition of Qwen3-30B-A3B by kalomaze.
Qwen3-30B-A3B converted to GGUF format
Qwen3-30B-A3B-Instruct-2507，Q8_0
这是unsloth的Q8动态量化版本，精度第一的量化版本！Unsloth Dynamic 2.0 实现了卓越的准确性，并超越了其他领先的量化模型。