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

Can I run gpt-oss-20b on my PC?

gpt-oss 20B (21B, 4.2B active) needs about 11.8 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 Arc A770 16GB (16 GB, $349 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 27.0–45.3 tokens/sec (usable for chat). The smallest Mac that runs it is the Mac M4 Pro 24GB.

11.8 GBneeded at 4-bit, 8K
36GPUs/Macs: fully in memory
15GPUs: with CPU offload
M4 Pro 24GBsmallest Mac

gpt-oss 20B 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 · ~27–45 tok/sdetails
RTX 40608 GBOffload Q4_K_M, 32 GB RAM · ~27–44 tok/sdetails
RTX 50608 GBOffload Q4_K_M, 32 GB RAM · ~29–52 tok/sdetails
RTX 3060 Ti8 GBOffload Q4_K_M, 32 GB RAM · ~29–52 tok/sdetails
RTX 4060 Ti 8GB8 GBOffload Q4_K_M, 32 GB RAM · ~27–45 tok/sdetails
RTX 30708 GBOffload Q4_K_M, 32 GB RAM · ~29–52 tok/sdetails
RTX 3080 10GB10 GBOffload Q4_K_M, 32 GB RAM · ~49–85 tok/sdetails
Arc B58012 GBOffload Q4_K_M, 32 GB RAM · ~72–98 tok/sdetails
RTX 3060 12GB12 GBOffload Q4_K_M, 32 GB RAM · ~63–83 tok/sdetails
RX 7700 XT12 GBOffload Q4_K_M, 32 GB RAM · ~70–94 tok/sdetails
RTX 507012 GBOffload Q4_K_M, 32 GB RAM · ~85–124 tok/sdetails
RTX 407012 GBOffload Q4_K_M, 32 GB RAM · ~75–104 tok/sdetails
RTX 4070 Super12 GBOffload Q4_K_M, 32 GB RAM · ~75–104 tok/sdetails
RTX 4070 Ti12 GBOffload Q4_K_M, 32 GB RAM · ~75–104 tok/sdetails
RTX 3080 12GB12 GBOffload Q4_K_M, 32 GB RAM · ~94–146 tok/sdetails
Arc A770 16GB16 GBQ8_0 · ~231 tok/sdetails
RTX 5060 Ti 16GB16 GBQ8_0 · ~185 tok/sdetails
RTX 4060 Ti 16GB16 GBQ8_0 · ~119 tok/sdetails
RX 7800 XT16 GBQ8_0 · ~257 tok/sdetails
RX 7900 GRE16 GBQ8_0 · ~237 tok/sdetails
RX 907016 GBQ8_0 · ~264 tok/sdetails
RX 9070 XT16 GBQ8_0 · ~264 tok/sdetails
RTX 5070 Ti16 GBQ8_0 · ~369 tok/sdetails
RTX 4070 Ti Super16 GBQ8_0 · ~277 tok/sdetails
RTX 4080 Super16 GBQ8_0 · ~303 tok/sdetails
RTX 508016 GBQ8_0 · ~395 tok/sdetails
RTX 408016 GBQ8_0 · ~295 tok/sdetails
RX 7900 XT20 GBF16 · ~289 tok/sdetails
RX 7900 XTX24 GBF16 · ~347 tok/sdetails
RTX 309024 GBF16 · ~338 tok/sdetails
RTX 409024 GBF16 · ~365 tok/sdetails
RTX 3090 Ti24 GBF16 · ~365 tok/sdetails
RTX 509032 GBF16 · ~648 tok/sdetails
DGX Spark128 GBF16 · ~99 tok/sdetails
Ryzen AI Max+ 395 128GB128 GBF16 · ~93 tok/sdetails
Mac M1 16GB16 GBNodetails
Mac M4 16GB16 GBNodetails
Mac M2 24GB24 GBF16 · ~36 tok/sdetails
Mac M3 24GB24 GBF16 · ~36 tok/sdetails
Mac M4 Pro 24GB24 GBF16 · ~99 tok/sdetails
Mac M4 32GB32 GBF16 · ~43 tok/sdetails
Mac M1 Pro 32GB32 GBF16 · ~72 tok/sdetails
Mac M3 Pro 36GB36 GBF16 · ~54 tok/sdetails
Mac M4 Pro 48GB48 GBF16 · ~99 tok/sdetails
Mac M4 Pro 64GB64 GBF16 · ~99 tok/sdetails
Mac M1 Max 64GB64 GBF16 · ~145 tok/sdetails
Mac M4 Max 64GB64 GBF16 · ~197 tok/sdetails
Mac M2 Max 96GB96 GBF16 · ~145 tok/sdetails
Mac M3 Max 128GB128 GBF16 · ~145 tok/sdetails
Mac M4 Max 128GB128 GBF16 · ~197 tok/sdetails
Mac M2 Ultra 192GB192 GBF16 · ~289 tok/sdetails
Mac M3 Ultra 256GB256 GBF16 · ~296 tok/sdetails
Mac M3 Ultra 512GB512 GBF16 · ~296 tok/sdetails

Questions

How much VRAM do I need for gpt-oss 20B?

About 11.8 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.

Why can a mixture-of-experts model run on a smaller card?

Only 4.2B of its 21B parameters are used per token. llama.cpp's --n-cpu-moe keeps the expert weights of some layers in system RAM while attention and the KV cache stay on the GPU, so each token reads far less from slow RAM than a dense model of the same size would.

Other popular checks: Can I run gpt-oss-120b?Can I run Llama 70B?Can I run Qwen3?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.