RunsOnMy ModelsGPUs & MacsBy VRAMBest GPUCan I run…?API pricesVRAM calculatorNew

Can I run Qwen3 on my PC?

Qwen3 32B (33B) needs about 21.1 GB of GPU memory at 4-bit with an 8K context. The cheapest consumer GPU in our table that runs it fully in VRAM at 4-bit or better is the RTX 5090 (32 GB, $1999 launch MSRP). With llama.cpp offload, the smallest card that can run it is the RX 7600 with 32 GB of system RAM, at an estimated 1.6–3.2 tokens/sec (slow but usable). The smallest Mac that runs it is the Mac M3 Pro 36GB.

21.1 GBneeded at 4-bit, 8K
14GPUs/Macs: fully in memory
28GPUs: with CPU offload
M3 Pro 36GBsmallest Mac

Every Qwen3 model we track

Cheapest consumer GPU (launch MSRP) and smallest Mac that run each model fully in memory at 4-bit or better with 8K context.

ModelParams4-bit · 8KCheapest GPUSmallest Mac
Qwen3 0.6B752M1.9 GBArc B580 ($249)Mac M4 16GB
Qwen3.5 0.8B873M1.5 GBArc B580 ($249)Mac M4 16GB
Qwen3 1.7B2.0B2.6 GBArc B580 ($249)Mac M4 16GB
Qwen3.5 2B2.3B2.2 GBArc B580 ($249)Mac M4 16GB
Qwen3 4B4.0B4.1 GBArc B580 ($249)Mac M4 16GB
Qwen3 4B Instruct 25074.0B4.1 GBArc B580 ($249)Mac M4 16GB
Qwen3 VL 4B Instruct4.4B4.1 GBArc B580 ($249)Mac M4 16GB
Qwen3.5 4B4.7B3.7 GBArc B580 ($249)Mac M4 16GB
Qwen3 8B8.2B6.5 GBArc B580 ($249)Mac M4 16GB
Qwen3 VL 8B Instruct8.8B6.5 GBArc B580 ($249)Mac M4 16GB
Qwen3.5 9B9.7B6.4 GBArc B580 ($249)Mac M4 16GB
Qwen3 14B15B10.3 GBArc A770 16GB ($349)Mac M4 Pro 24GB
Qwen3.8 27B28B16.8 GBRX 7900 XT ($899)Mac M1 Pro 32GB
Qwen3.6 27B28B17.0 GBRX 7900 XT ($899)Mac M1 Pro 32GB
Qwen3.5 27B28B17.0 GBRX 7900 XT ($899)Mac M1 Pro 32GB
Qwen3 Coder 30B A3B Instruct31B MoE18.7 GBRX 7900 XTX ($999)Mac M1 Pro 32GB
Qwen3 32B33B21.1 GBRTX 5090 ($1999)Mac M3 Pro 36GB
Qwen3.6 35B A3B36B MoE22.1 GBRTX 5090 ($1999)Mac M4 Pro 48GB
Qwen3.5 35B A3B36B MoE21.5 GBRTX 5090 ($1999)Mac M3 Pro 36GB
Qwen3.8 Flash Next180B MoE103.5 GBnone in VRAMMac M2 Ultra 192GB
Qwen3.8 2.4T A95B2.45T MoE1394.1 GBnone in VRAMnone

Qwen3 32B on every GPU and Mac

Best quant that fits fully in memory at 8K context with a speed ceiling; otherwise the llama.cpp offload plan (system RAM needed, estimated speed from DDR4-3200 to DDR5-6000). Macs and unified-memory PCs share one pool, so offload adds nothing there. How offload is estimated.

HardwareMemoryVerdictDetails
RX 76008 GBOffload Q4_K_M, 32 GB RAM · ~1.6–3.2 tok/sdetails
RTX 40608 GBOffload Q4_K_M, 32 GB RAM · ~1.6–3.2 tok/sdetails
RTX 50608 GBOffload Q4_K_M, 32 GB RAM · ~1.6–3.3 tok/sdetails
RTX 3060 Ti8 GBOffload Q4_K_M, 32 GB RAM · ~1.6–3.3 tok/sdetails
RTX 4060 Ti 8GB8 GBOffload Q4_K_M, 32 GB RAM · ~1.6–3.2 tok/sdetails
RTX 30708 GBOffload Q4_K_M, 32 GB RAM · ~1.6–3.3 tok/sdetails
RTX 3080 10GB10 GBOffload Q4_K_M, 32 GB RAM · ~1.9–3.9 tok/sdetails
Arc B58012 GBOffload Q4_K_M, 32 GB RAM · ~2.2–4.2 tok/sdetails
RTX 3060 12GB12 GBOffload Q4_K_M, 32 GB RAM · ~2.1–4.1 tok/sdetails
RX 7700 XT12 GBOffload Q4_K_M, 32 GB RAM · ~2.2–4.2 tok/sdetails
RTX 507012 GBOffload Q4_K_M, 32 GB RAM · ~2.2–4.4 tok/sdetails
RTX 407012 GBOffload Q4_K_M, 32 GB RAM · ~2.2–4.3 tok/sdetails
RTX 4070 Super12 GBOffload Q4_K_M, 32 GB RAM · ~2.2–4.3 tok/sdetails
RTX 4070 Ti12 GBOffload Q4_K_M, 32 GB RAM · ~2.2–4.3 tok/sdetails
RTX 3080 12GB12 GBOffload Q4_K_M, 32 GB RAM · ~2.3–4.5 tok/sdetails
Arc A770 16GB16 GBOffload Q4_K_M, 32 GB RAM · ~3.2–5.9 tok/sdetails
RTX 5060 Ti 16GB16 GBOffload Q4_K_M, 32 GB RAM · ~3.1–5.6 tok/sdetails
RTX 4060 Ti 16GB16 GBOffload Q4_K_M, 32 GB RAM · ~2.9–5.0 tok/sdetails
RX 7800 XT16 GBOffload Q4_K_M, 32 GB RAM · ~3.2–6.0 tok/sdetails
RX 7900 GRE16 GBOffload Q4_K_M, 32 GB RAM · ~3.2–6.0 tok/sdetails
RX 907016 GBOffload Q4_K_M, 32 GB RAM · ~3.2–6.1 tok/sdetails
RX 9070 XT16 GBOffload Q4_K_M, 32 GB RAM · ~3.2–6.1 tok/sdetails
RTX 5070 Ti16 GBOffload Q4_K_M, 32 GB RAM · ~3.3–6.4 tok/sdetails
RTX 4070 Ti Super16 GBOffload Q4_K_M, 32 GB RAM · ~3.2–6.1 tok/sdetails
RTX 4080 Super16 GBOffload Q4_K_M, 32 GB RAM · ~3.3–6.2 tok/sdetails
RTX 508016 GBOffload Q4_K_M, 32 GB RAM · ~3.3–6.5 tok/sdetails
RTX 408016 GBOffload Q4_K_M, 32 GB RAM · ~3.2–6.2 tok/sdetails
RX 7900 XT20 GBOffload Q4_K_M, 32 GB RAM · ~6.6–11 tok/sdetails
RX 7900 XTX24 GBQ3_K_M only (below 4-bit)details
RTX 309024 GBQ3_K_M only (below 4-bit)details
RTX 409024 GBQ3_K_M only (below 4-bit)details
RTX 3090 Ti24 GBQ3_K_M only (below 4-bit)details
RTX 509032 GBQ5_K_M · ~77 tok/sdetails
DGX Spark128 GBF16 · ~4 tok/sdetails
Ryzen AI Max+ 395 128GB128 GBF16 · ~4 tok/sdetails
Mac M1 16GB16 GBNodetails
Mac M4 16GB16 GBNodetails
Mac M2 24GB24 GBQ2_K only (below 4-bit)details
Mac M3 24GB24 GBQ2_K only (below 4-bit)details
Mac M4 Pro 24GB24 GBQ2_K only (below 4-bit)details
Mac M4 32GB32 GBQ3_K_M only (below 4-bit)details
Mac M1 Pro 32GB32 GBQ3_K_M only (below 4-bit)details
Mac M3 Pro 36GB36 GBQ4_K_M · ~8 tok/sdetails
Mac M4 Pro 48GB48 GBQ6_K · ~10 tok/sdetails
Mac M4 Pro 64GB64 GBQ8_0 · ~8 tok/sdetails
Mac M1 Max 64GB64 GBQ8_0 · ~11 tok/sdetails
Mac M4 Max 64GB64 GBQ8_0 · ~16 tok/sdetails
Mac M2 Max 96GB96 GBF16 · ~6 tok/sdetails
Mac M3 Max 128GB128 GBF16 · ~6 tok/sdetails
Mac M4 Max 128GB128 GBF16 · ~8 tok/sdetails
Mac M2 Ultra 192GB192 GBF16 · ~12 tok/sdetails
Mac M3 Ultra 256GB256 GBF16 · ~12 tok/sdetails
Mac M3 Ultra 512GB512 GBF16 · ~12 tok/sdetails

Questions

How much VRAM do I need for Qwen3 32B?

About 21.1 GB at 4-bit (Q4_K_M) with an 8K context, including KV cache and runtime overhead. Longer contexts need more; the model page has the full table by quant and context.

Other popular checks: Can I run gpt-oss-120b?Can I run gpt-oss-20b?Can I run Llama 70B?Can I run DeepSeek R1? · Best GPU by budget · Mac vs PC

Model data: Hugging Face Hub API and GGUF file sizes. GPU specs: vendor pages. Speeds are estimates, not benchmarks. Methodology · Report an error.