Models / Qwen3-Coder 30B-A3B
Best home rig for Qwen3-Coder 30B-A3B
Quick answer
How much VRAM does Qwen3-Coder 30B-A3B need?
The weights are about 18.7 GB at Q4_K_M (30.5B params × 4.89 bits ÷ 8, before context/KV cache). Add memory for context (KV cache) and the runtime.
What is the fastest measured hardware for Qwen3-Coder 30B-A3B?
RX 9070 16 GB on Windows (HIP): Qwen3-Coder 30B-A3B Q3_K_L at 83.3 tok/s with an 8K prompt, 50.1 at 32K: 83.3 tok/s generation (llama.cpp b11146 (HIP, ROCm 10.0.0), Q3_K_L, q4_0 KV), from https://minjunpark.blog/posts/windows-rx9070-local-llm-benchmarks/.
Last updated:
Legacy model30.5B total · 3.3B active2 measured resultsfastest plain decode: 83.3 tok/s model info ↗
Older model. For current picks see current models.
Real, measured generation speeds for Qwen3-Coder 30B-A3B on home hardware, fastest first. Different quants/engines aren't directly comparable.
| Build | Quant | Engine | tok/s | Prompt processing (tok/s) | Cost | Source |
|---|---|---|---|---|---|---|
| RX 9070 16 GB on Windows (HIP): Qwen3-Coder 30B-A3B Q3_K_L at 83.3 tok/s with an 8K prompt, 50.1 at 32K | Q3_K_L, q4_0 KV | llama.cpp b11146 (HIP, ROCm 10.0.0) | 83.3 | 1,802 @8K · est. TTFT @8k ≈4.5 s* | minjunpark.blog ↗ | |
| RX 9070 16 GB on Windows (HIP): Qwen3-Coder 30B-A3B Q3_K_L at 83.3 tok/s with an 8K prompt, 50.1 at 32K | Q3_K_L, q4_0 KV | llama.cpp b11146 (HIP, ROCm 10.0.0) | 50.1 | 1,790 @32K · est. TTFT @8k ≈4.6 s* | minjunpark.blog ↗ |
Prompt processing = prefill speed reported by the source, @ prompt length. * est. TTFT @8k is calculated from measured prefill, not measured. Why it matters.
Memory math: how much VRAM / RAM does Qwen3-Coder 30B-A3B need?
Qwen3-Coder 30B-A3B has 30.5B parameters, about 3.3B active per token (source: Qwen model card: 30.5B total / 3.3B activated MoE coding model (2025)). Weight size = parameters × bits-per-weight ÷ 8, before context (KV cache) and runtime overhead.
| Format | Bits/weight | Weights | Smallest common memory size that holds the weights alone |
|---|---|---|---|
| Q8_0 | 8.50 | 32.4 GB | 48 GB |
| Q6_K | 6.56 | 25.0 GB | 32 GB |
| Q5_K_M | 5.70 | 21.7 GB | 24 GB |
| Q4_K_M | 4.89 | 18.7 GB | 24 GB |
| Q3_K_M | 4.00 | 15.2 GB | 16 GB |
| Q2_K | 3.16 | 12.0 GB | 16 GB |
Bits-per-weight for GGUF quants are llama.cpp's own figures (llama.cpp quantize README). "As measured" rows apply the source's stated bits-per-weight to the parameter count; real files differ slightly because some layers stay at higher precision. Try other combinations in the VRAM calculator.
Why it can still run on smaller GPUs: it is a mixture-of-experts model, so only about 3.3B of its 30.5B parameters are used for each generated token. Runtimes such as llama.cpp (--n-cpu-moe), ExLlamaV3 or Project Maya can keep most expert weights in system RAM and only the shared layers and hot experts in VRAM. Then you need VRAM + system RAM ≥ weights, and speed is limited mainly by RAM bandwidth.
Measured on setups with less GPU memory than the Q4_K_M weights (so part of the model lived in system RAM or was streamed from disk):
- RX 9070 16GB (16 GB) + about 32 GB: 83.3 tok/s (Q3_K_L, q4_0 KV)
- RX 9070 16GB (16 GB) + about 32 GB: 50.1 tok/s (Q3_K_L, q4_0 KV)
FAQ
How much VRAM does Qwen3-Coder 30B-A3B need?
The weights are about 18.7 GB at Q4_K_M (30.5B params × 4.89 bits ÷ 8, before context/KV cache). Add memory for context (KV cache) and the runtime.
What is the fastest measured hardware for Qwen3-Coder 30B-A3B?
RX 9070 16 GB on Windows (HIP): Qwen3-Coder 30B-A3B Q3_K_L at 83.3 tok/s with an 8K prompt, 50.1 at 32K: 83.3 tok/s generation (llama.cpp b11146 (HIP, ROCm 10.0.0), Q3_K_L, q4_0 KV), from https://minjunpark.blog/posts/windows-rx9070-local-llm-benchmarks/.
Related: what GPU do you need for 2026 models · all builds