作者： Jakub Rusinowski · 最后更新： 2026年6月26日
PREVIEW (June 2026, specs unverified). Alibaba's follow-up to Qwen 3.6 — a 35B-A3B MoE with native multimodal input, Apache 2.0, and ~256K context (extensible). Consumer-runnable at Q4. Numbers carried from the Qwen 3.6 lineage; verify against the Hugging Face model card.
| Licence | What it permits | Applies to |
|---|---|---|
Apache-2.0 | Commercial use permitted Commercial use permitted. No usage restrictions beyond attribution. | Qwen 3.7 35B-A3B |
| Qwen 3.7 35B-A3B | Min 22 GB VRAM · Q4_K_M · 262,144 ctx · |
The cheapest GPU that runs Qwen 3.7 locally (min 22 GB VRAM) is the AMD Radeon RX 7900 XTX (24 GB).
Install Ollama then run: ollama run
Minimum VRAM: 22 GB. For best results use Q4_K_M quantization.
Qwen 3.7 needs about 22 GB VRAM at Q4_K_M quantization for its smallest variant. Variants: Qwen 3.7 35B-A3B (22 GB, Q4_K_M). On Apple Silicon, unified memory counts toward this requirement.
Yes — Qwen 3.7 runs on an RTX 4090 (24 GB) and other 24 GB cards such as the RTX 3090. Smaller variants also fit comfortably on 8–16 GB GPUs at Q4_K_M.
Q4_K_M is the best balance of quality and VRAM for Qwen 3.7 in most cases. Choose Q8_0 for near-lossless quality if you have spare VRAM, or smaller quants (Q3/Q2) only when memory is tight.
Install Ollama, then run: ollama run . This downloads Qwen 3.7 and starts a local, OpenAI-compatible endpoint — no internet connection is needed after the initial download.