← AI Hashrate

Can L40S 48GB run MiniCPM5-1B?

NVIDIA · 48 GB GDDR6 · 864 GB/s bandwidth · OpenBMB · 1.08B · model ctx up to 128K

Yes — the L40S 48GB (48 GB GDDR6) can run MiniCPM5-1B. At Q4 with the default 8K context it needs ≈ 1.81 GB VRAM (weights 0.59 GB + KV cache 0.22 GB + 1 GB runtime overhead), which fits within 48 GB. Decode at Q4/8K: ≈ 509.1 tok/s (estimated, batch 1). FP16 (≈ 3.38 GB) also fits, at ≈ 140.0 tok/s (estimated). Estimates come from memory-bandwidth math; rows tagged measured override estimates. Relative ranking is more reliable than absolute tok/s. Methodology.

Fit & speed by quant and context

QuantContextVRAM neededFits?tok/s (decode)
Q44K1.7 GBYes509.1 est.
Q48K default1.81 GBYes509.1 est.
Q432K2.46 GBYes509.1 est.
FP164K3.27 GBYes140.0 est.
FP168K default3.38 GBYes140.0 est.
FP1632K4.02 GBYes140.0 est.

Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s means the model is far beyond this card (offload-only territory).

VRAM breakdown at 8K context

QuantWeightsKV cacheOverheadTotal neededL40S 48GB VRAM
Q40.59 GB0.22 GB1 GB1.81 GB48 GB
FP162.16 GB0.22 GB1 GB3.38 GB48 GB

Other GPUs that run MiniCPM5-1B

All GPUs for MiniCPM5-1B →

Other models for the L40S 48GB

All models on the L40S 48GB →