MiniCPM5-1B

OpenBMB · 1.08B · ctx 128K · apache-2.0 · HuggingFace

MiniCPM5-1B (OpenBMB) is a 1.08B parameters open-weight model tracked on AI Hashrate. At 8K context we estimate roughly 1.81 GB VRAM for Q4 (weights 0.59 GB + KV 0.22 GB + 1 GB overhead) and 3.38 GB for FP16. About 50 GPUs/accelerators in our catalog fully fit this model at Q4 under that context assumption. Fastest Q4 configs: B200 192GB SXM ≈ 4713.8 tok/s (estimated); MI300X 192GB ≈ 3122.9 tok/s (estimated); H200 141GB SXM5 ≈ 2828.3 tok/s (estimated); H100 80GB SXM5 ≈ 1973.9 tok/s (estimated); Gaudi 3 128GB ≈ 1930.3 tok/s (estimated). Use the table below for VRAM fit and tok/s per dollar (list MSRP). Estimates follow memory-bandwidth math; measured rows override when present. See methodology for details. Methodology.

100 hardware configs — sorted by tok/s. Default fit context 8K.

Fit checks

• RTX 5090 32GB — fits · ≈1055.9 tok/s Q4 est.
• RTX 3090 Ti 24GB — fits · ≈593.9 tok/s Q4 est.
• RTX 4090 24GB — fits · ≈593.9 tok/s Q4 est.
• RTX 5080 16GB — fits · ≈565.7 tok/s Q4 est.
• RTX 6000 Ada 48GB — fits · ≈565.7 tok/s Q4 est.
• RX 7900 XTX 24GB — fits · ≈565.7 tok/s Q4 est.
• RTX 3090 24GB — fits · ≈551.5 tok/s Q4 est.
• RTX 5070 Ti 16GB — fits · ≈527.9 tok/s Q4 est.
• L40S 48GB — fits · ≈509.1 tok/s Q4 est.
• Mac Studio M3 Ultra 96GB — fits · ≈482.6 tok/s Q4 est.
• Mac Studio M2 Ultra 192GB — fits · ≈471.4 tok/s Q4 est.
• RX 7900 XT 20GB — fits · ≈471.4 tok/s Q4 est.

12 popular retail GPUs; every row in the table above also links its Fits? verdict to the full fit check.

Same family: MiniCPM

All MiniCPM models →