llamaperf

NVIDIA B300 288GB vs NVIDIA RTX 3090 for local LLMs

3 reports on the NVIDIA B300 288GB and 271 on the NVIDIA RTX 3090

Which is faster for local LLMs?

Nobody has reported the same model at the same quant on both cards yet, so there's no measured head-to-head. Memory bandwidth, which caps how fast a card can write, is 8.5x higher on the NVIDIA B300 288GB (8,000 GB/s on the NVIDIA B300 288GB, 936 on the NVIDIA RTX 3090). The estimates below compare them on the models people run most.

NVIDIA B300 288GB
Memory
288GB
Memory bandwidth
8,000 GB/s
FP16 compute
2250 TFLOPS
Reports
3
NVIDIA RTX 3090
Memory
24GB
Memory bandwidth
936 GB/s
FP16 compute
142 TFLOPS
Reports
271

Estimated by the calculator

The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape.

ModelNVIDIA B300 288GBNVIDIA RTX 3090
Qwen3.8 27B81.5 t/sFP1633.9 t/sQ4_K_M
Qwen3.8 125B · 6B active293.3 t/sQ8_019.4 t/sQ2_K, experts in RAM
Qwen3.6 35B · 3B active293.3 t/sFP16120.1 t/sQ4_K_S
DeepSeek V4 Flash 284B · 13B active264.1 t/sQ6_Kdoesn't fit at these settings
DeepSeek V4.1 Flash 552B · 16B active379.5 t/sQ3_K_Mdoesn't fit at these settings
Qwen3.6 27B81.5 t/sFP1633.9 t/sQ4_K_M

Change the context, RAM or model in the calculator: NVIDIA B300 288GB · NVIDIA RTX 3090

Every report on each card, with its source: NVIDIA B300 288GB · NVIDIA RTX 3090

More comparisons