Can you run MiniMaxAI/MiniMax-M1-40k on RTX 3060 12GB?

MiniMaxAI/MiniMax-M1-40k needs ~986.26 GB of memory (weights + KV cache at 8192 tokens); that is more than the 12 GB on RTX 3060 12GB, so it does not fit. Computed from stated size and quant, not a runtime guarantee.

MiniMaxAI/MiniMax-M1-40k needs ~986.26 GB of memory (weights + KV cache at 8192 tokens); that is more than the 12 GB on RTX 3060 12GB, so it does not fit. Computed from stated size and quant, not a runtime guarantee.

Memory-capacity estimate across common GPUs

GPUVerdictVRAMNeeds
8GB laptop (no dedicated GPU)Does not fit8 GB~986.26 GB
16GB laptop (no dedicated GPU)Does not fit16 GB~986.26 GB
NVIDIA T4 16GB (free Colab)Does not fit16 GB~986.26 GB
NVIDIA L4 24GBDoes not fit24 GB~986.26 GB
RTX 3060 12GBDoes not fit12 GB~986.26 GB
RTX 4080 16GBDoes not fit16 GB~986.26 GB
RTX 3090 24GBDoes not fit24 GB~986.26 GB
RTX 4090 24GBDoes not fit24 GB~986.26 GB
A100 40GBDoes not fit40 GB~986.26 GB
A100 80GBDoes not fit80 GB~986.26 GB
H100 80GBDoes not fit80 GB~986.26 GB
Apple M-series 16GB (unified)Does not fit16 GB~986.26 GB
Apple M-series 32GB (unified)Does not fit32 GB~986.26 GB
Apple M-series 64GB (unified)Does not fit64 GB~986.26 GB
Apple M-series 128GB (unified)Does not fit128 GB~986.26 GB
CPU / 32GB system RAMDoes not fit32 GB~986.26 GB
CPU / 64GB system RAMDoes not fit64 GB~986.26 GB

How this was computed

  • Weights + KV cache at the stated context + a runtime/overhead allowance; memory capacity only, not a benchmark.

Next steps

Methodology: weights + KV-cache at 8192 tokens + a runtime overhead allowance, from the recommended download size, parameter count, and quant. Computed from stated size and quant, not a runtime guarantee. Not a benchmark; runtimes differ.