OpenAI adds economic model to Dots agents
OpenAI announced on November 3, 2026, the addition of the smaller and less expensive GPT-6.1 Terra model to power part of the tasks for its Dots personal assistants via an automatic router to reduce cloud computing consumption. This comes amid funding pressures and heavy computing costs facing the persistent agent sector.
OpenAI challenges Nvidia with groundbreaking inference chip
OpenAI has published benchmark results proving that its custom inference chip, “Jalapeno,” outperforms Nvidia’s leading systems in energy efficiency, throughput, and latency. Developed in collaboration with Broadcom, the chip represents the first public and direct challenge to Nvidia’s absolute dominance of the global AI hardware market.