TEAS benchmark: 5 models on 9 accelerators across 6 realistic agentic workloads
PontiEdoardo · x · 2026-09-01
ARIA Research's Scaling Compute programme released TEAS 🍵, a benchmark serving 5 models (4B to 1T total params) on 9 accelerators across 6 workloads, arguing next-gen accelerators must be benchmarked on realistic agentic workloads. TEAS profiles workloads by bottleneck (prefill, decode, tool use) and user choices (stack, batch regime, budget), highlighting each accelerator and stack's strengths per scenario instead of a one-dimensional ranking.
More from Infra
- DIT launches AI token exchange to route requests, claiming 30–70% cost savings — Div_pradeep · 2026-09-01
- Anthropic commits $80B to cloud capacity in a single month — kimmonismus · 2026-09-01
- ESP32 voice assistant: 8x wake-word model compression with AIMET — carrycooldude · 2026-09-01
- LITE plans VCSEL products for AI interconnects, delayed by 1-2 years — zephyr_z9 · 2026-09-01
- Xiaohongshu & NVIDIA build GR-Inference engine, doubling throughput for Beam Search — 小红书技术REDtech · 2026-09-01
- antirez shows DeepSeek v4 Flash vision running fast locally on an M5 Max; Metal/CUDA/ROCm support nearly ready — antirez · 2026-09-01