Koyeb GPU LLM Benchmarks (vLLM benchmarking cli)

Fields

id

sources/koyeb-gpu-benchmarks-vllm-cli

URL

https://www.koyeb.com/docs/hardware/gpu-benchmarks

Publisher

Koyeb

Kind

benchmark

Accessed

2026-10-15

notes

Vendor-hosted third-party measurement. vLLM benchmarking cli with random data, token shapes 512x512/1024x1024/4096x1024, batch 1/8/32. Throughput t/s = total tokens/wall-clock (input+output), AGGREGATE over batch. Llama-3.1-8B, DeepSeek-R1-Llama-8B, Qwen2.5-7B-Instruct on A100(PCIe)/A100SXM/H100/H200/L40S/RTX-PRO-6000. Last updated 2026-05-27.

Notes

Vendor-hosted third-party measurement. vLLM benchmarking cli with random data, token shapes 512x512/1024x1024/4096x1024, batch 1/8/32. Throughput t/s = total tokens/wall-clock (input+output), AGGREGATE over batch. Llama-3.1-8B, DeepSeek-R1-Llama-8B, Qwen2.5-7B-Instruct on A100(PCIe)/A100SXM/H100/H200/L40S/RTX-PRO-6000. Last updated 2026-05-27.

Cross-references

This record references no other records.