UK Universities Open-Source TEAS Benchmark for AI Inference Across 9 Accelerators
Edinburgh, EPCC and Imperial College have open-sourced TEASBench under ARIA's Scaling Compute program, testing 5 models across 9 accelerators on real agent workloads. It measures inference at natural dataset lengths and reports cost and energy consumption.
2026-09-01 ~ 2026-09-01 · 3 related posts
- TEAS benchmark: 5 models on 9 accelerators across 6 realistic agentic workloads — PontiEdoardo · 2026-09-01
- TEAS benchmark: measures inference at natural lengths, reporting cost, accuracy, and energy — PontiEdoardo · 2026-09-01
- UK universities release TEASBench: measuring AI inference cost and performance across hardware — PontiEdoardo · 2026-09-01