Hands-Free Coding and Faster Inference Claims

OpenAI has introduced desktop voice control for Codex and ChatGPT, while AMD and Cerebras have announced an inference solution targeting ultra-low latency and high throughput. The developments concern how developers interact with AI and how applications serve it.

Voice Control Changes the Coding Interface

OpenAI has introduced full duplex voice control for Codex and ChatGPT on desktop, enabling hands-free coding. That adds a way to interact with coding tools; the announcement does not establish a productivity gain. Separately, The New Stack examines whether expensive reasoning models are always worth it for coding, raising a model-selection question alongside the interface change.

Inference Announcement Targets Latency and Throughput

AMD and Cerebras have announced a joint AI inference solution that they describe as offering industry-leading performance, with ultra-low latency and high throughput. Those are the companies' performance promises. The announcement targets applications requiring rapid processing and responses, but the available material provides no measurements for comparing the solution with alternatives.