LLMLocal inference
Qwen3 235B-A22B
Weight estimates and planning VRAM for running Qwen3 235B-A22B locally at each quantization level. Compare the lowest reference-cost devices that clear the plan.
Q4 plan
152 GB
Q8 plan
276 GB
FP16 plan
541 GB
Context window
128 k tokens
Planning targets add 15% to weight estimates for runtime buffers and a modest KV cache. Long context or a different backend can require more. Official Qwen model card ↗
01 // GPUs that can run Qwen3 235B-A22B
Compatible hardware by quantization
Sorted by dated reference-cost estimates from Apr 30, 2026. These are not live offers.
Q4Q4_K_M (4-bit)
132GB weights · plan ≥152GBGPUVRAMPriceTier
Q8Q8_0 (8-bit)
240GB weights · plan ≥276GBGPUVRAMPriceTier
FP16FP16 (full precision)
470GB weights · plan ≥541GBNo single GPU tracked here clears the 541GB planning target for Qwen3 235B-A22B at FP16. Multi-GPU, offload, or cloud inference is needed at this precision.
02 // Frequently asked
Qwen3 235B-A22B GPU questions
How much VRAM does Qwen3 235B-A22B need?
Qwen3 235B-A22B uses approximately 132GB for Q4 weights, 240GB at Q8, or 470GB at FP16. GPU Hunter adds 15% planning headroom for runtime buffers and a modest KV cache, producing targets of 152GB, 276GB, and 541GB respectively. Exact memory use varies by backend and context length.
What is the cheapest GPU to run Qwen3 235B-A22B?
Using GPU Hunter's 152GB Q4 planning target, the lowest reference-cost single device is the Apple M3 Ultra (512GB VRAM, dated estimate $9,499).
Can I run Qwen3 235B-A22B at FP16?
Qwen3 235B-A22B uses about 470GB for FP16 weights and 541GB under GPU Hunter's planning allowance—well beyond a single consumer GPU. Q4 or Q8 is more practical.
What quantization is best for Qwen3 235B-A22B?
Q4_K_M uses about 132GB for weights and is the most hardware-accessible option. Q8_0 uses about 240GB and trades more memory for fidelity. FP16 uses about 470GB before runtime and context overhead. The right choice depends on the task, backend, and context window.