largeApache 2.030.5B paramsOpen weights

Qwen3-Coder 30B (A3B)

30.5B parameter open-weight model. The coding variant of the 30B sparse MoE - same 128-expert, ~3B-active architecture tuned for agentic coding, with native 256K context for large repos. Q4_K_M is ~19 GB; measured 31.4 tok/s on an RTX 3080 (10 GB) at 46% VRAM residency (2026-08-01), so it stays responsive even when split. Apache 2.0. The natural upgrade path from Qwen2.5-Coder.

Alibaba · Qwen3 Coder

Editorial details

EditorCJ ZwartLast updatedOctober 2026SourcesHuggingFace model card (Qwen/Qwen3-Coder-30B-A3B-Instruct) and official docs.

VRAM figures are empirical estimates. Actual usage varies by runtime, context length, and system configuration. Verify on your specific hardware before production use.

Will Qwen3-Coder 30B (A3B) run on your machine?

Qwen3-Coder 30B (A3B) is 30.5B parameters and needs 20.1 GB of VRAM at Q4_K_M: 18.6 GB of weights plus 1.5 GB of runtime overhead for the inference server itself.

VRAM by quantization

QuantizationWeightsNeeds (with overhead)Quality
Q4_K_M18.6 GB20.1 GBgood

Fit on common hardware at Q4_K_M

HardwareMemory the model can useSystem RAMVerdict
CPU OnlyNone (CPU only)16 GBToo large
RTX 4060 Laptop8 GB16 GBToo large
RTX 3060 (12GB)12 GB32 GBCPU offload
RTX 4060 Ti (16GB)16 GB32 GBCPU offload
RTX 309024 GB64 GBComfortable
Apple Silicon (Unified Memory) 36 GB24 GB of 36 GB36 GBComfortable
RTX 509032 GB64 GBComfortable

Comfortable means VRAM clears the requirement by 2 GB or more. Tight means it covers the requirement with no margin. CPU offload means the model does not fit in VRAM but system RAM is at least 1.6× the weights, so it will run at reduced speed: expect roughly 1–5 tokens per second. Figures are weights plus a fixed runtime overhead and exclude KV-cache growth, which scales with context length.

Apple Silicon shares one pool of memory between the system and the GPU, so a model cannot use all of it. These rows apply the same usable share the Compatibility Checker uses (two-thirds of the pool on Macs up to 36 GB, three-quarters above), which is why a 36 GB Mac is graded on 24 GB.

Check Qwen3-Coder 30B (A3B) against your exact hardwareHow these verdicts are calculated

Want a broader hardware plan around Qwen3-Coder 30B (A3B)? Open the PC Builder for the whole 30B / 32B class → The Builder plans across that class; it does not replace the exact-model fit verdict above.

Related models

North Mini Code 1.030.5B parameter model, shared focus: coding, agents, tool-use.Qwen3.8 27B27.8B parameter model, shared focus: chat, reasoning, long-context.Laguna XS 2.133.4B parameter model, shared focus: coding, agents, tool-use.Qwen2.5 14BBuilders hosting local AI on a single RTX 4080 (16 GB) or RTX 4090 (24 GB) who want near-frontier multilingual reasoning, mathematical accuracy, and agentic coding performance without multi-GPU infrastructure. The sweet spot between consumer accessibility and frontier-class open-weight capability for 2026 local deployment.

Ready to run this model locally?

Find a compatible interface in our Local AI Tools directory →