RunsLocalGPU checkerLocal AI toolsv1.9 · data Oct 15, 2026

All models / Nemotron 3.5 Lightning 30B-A3B

What GPU do I need to run Nemotron 3.5 Lightning 30B-A3B?

Nemotron 3.5 Lightning 30B-A3B has 30B parameters (3B active per token, mixture-of-experts). At Q4_K_M with 8K context it needs about 19.0 GB of memory. The smallest graphics card that runs it fully on the GPU is about 20 GB, such as the RX 7900 XT. Because only 3B parameters are active, it still runs usably with offload to system RAM on cards like the RTX 5080.

Check your own hardware

Memory by quantization (8K context)

QuantBits per weightMemory needed
Q8_08.532.7 GB
Q6_K6.5625.4 GB
Q5_K_M5.6922.1 GB
Q4_K_M4.8519.0 GB
Q3_K_M3.9115.5 GB
IQ2_M2.710.9 GB

Hardware that runs it fully in fast memory

HardwareMemoryBest quantEst. speed
RX 7900 XT20 GBQ4_K_M150+ tok/s
RTX 409024 GBQ5_K_M150+ tok/s
RTX 3090 Ti24 GBQ5_K_M150+ tok/s
RX 7900 XTX24 GBQ5_K_M150+ tok/s
RTX 309024 GBQ5_K_M150+ tok/s
RTX 509032 GBQ6_K150+ tok/s
M5 32GB32 GBQ5_K_M43 tok/s
M4 32GB32 GBQ5_K_M34 tok/s
M4 Max 36GB36 GBQ6_K100 tok/s
M4 Pro 48GB48 GBQ8_051 tok/s
M4 Max 64GB64 GBQ8_0103 tok/s
Ryzen AI Max+ 395 64GB64 GBQ8_048 tok/s
RTX PRO 6000 Blackwell96 GBQ8_0150+ tok/s
M3 Ultra 96GB96 GBQ8_0150+ tok/s
M4 Max 128GB128 GBQ8_0103 tok/s
DGX Spark 128GB128 GBQ8_051 tok/s
Ryzen AI Max+ 395 128GB128 GBQ8_048 tok/s
M3 Ultra 256GB256 GBQ8_0150+ tok/s
M3 Ultra 512GB512 GBQ8_0150+ tok/s

Runs with offload to system RAM (slower)

Assumes 32 GB of DDR5 system RAM for offload. Estimates only; real speeds usually land within about 20%.