Agents
Google unveils two new TPUs designed for the 'agentic era'
Google unveiled two eighth-generation Tensor Processing Units—the TPU 8t for training and TPU 8i for inference—designed with specialized architectures for AI model training and agent development. The
Google unveiled two eighth-generation Tensor Processing Units—the TPU 8t for training and TPU 8i for inference—designed with specialized architectures for AI model training and agent development. The TPU 8i is built to handle collaborative agent workloads with increased memory bandwidth for latency-sensitive inference, scaling to 1,152 chips in a single pod. These chips represent Google's strategic shift to help enterprises move from merely using AI chatbots to building full-blown agentic AI systems.
Related
- Come check out GLM 5.1 in Code Arena for agentic web development tasks using tools. Don’t forget to vote, Code Arena scores are coming up ne…
- GLM-5.1 is now available in Windsurf! Try it out and let us know what you think
- GLM-5.1 gives teams a stronger model for coding, tool use, and sustained agent performance on Together AI. Learn more: http://www.together.a…
- We are entering an extremely exciting era for open-weight models. Kimi K2.6 now feels like a top agentic model. I took it for a spin via @Fi…
Source: Ars Technica | 2026-04-22