Crusoe reports inference results on 512 AMD MI355X GPUs
In a September 16, 2026 publication, Crusoe reports MLPerf Inference v6.1 results: 5.75 million tokens per second on gpt-oss-120b and linear scaling without InfiniBand.
Source: Crusoe relata resultados de inferência do MLPerf em AMD MI355X (crusoe.ai). Text prepared with AI from this source.
What happened and what to do
Crusoe's post, dated September 16, 2026, describes MLPerf Inference v6.1 results using 512 AMD Instinct MI355X GPUs. The company reports 5.75 million tokens per second on the gpt-oss-120b workload and linear scaling without InfiniBand. The figures offer a reference for assessing throughput and scalability in inference clusters.
A company planning to expand inference capacity can benchmark its own cluster with representative workloads, monitor throughput and latency, and identify network, hardware, and operational constraints. A collection pipeline and dashboard can consolidate these measurements to inform capacity, architecture, and cost decisions, without assuming another environment's results will recur.
How the consultancy can help
Wendelmaques can diagnose opportunities or bottlenecks in your environment, define metrics and test workloads, and scope an implementation for data collection, monitoring, and dashboards, with operational support.
Next step
Send a short description of your environment or inference challenge to receive a scoped proposal.
Let's assess your project
I reply with scope and a proposal.