verifiedhigh · 0.85updated 2026-10-14

MLPerf v6.1 Oracle 8x GB300: llama2-70b-99.9 Server aggregate tok/s

127983.87 tok/s AGGREGATE over 8x GB300

Fields

benchmarks/mlperf-v6-1-oracle-8xgb300-llama2-70b-server-tokens

Metric

tps_aggregate

Value

127,983.87

Unit

tok/s AGGREGATE over 8x GB300

Model

llama2-70b-99.9

Methodology

MLPerf Inference closed-division submission. v6.1, Oracle 8x GB300. Server: 'Completed tokens per second: 127983.87'. Values quoted verbatim from the submission's mlperf_log_summary.txt (performance/run_1). AGGREGATE throughput across all accelerators in the system, NOT per-request.

Measured by

vendor

Reproducible

yes

Cross-references

MLPerf v6.1 Oracle 8x GB300: llama2-70b-99.9 Server aggregate tok/s

Fields

Cross-references

Accelerators (1)

Engine (1)

Interconnect (2)