OpenAI adds voice agent mode to Codex

// 2h agoPRODUCT UPDATE

OpenAI adds voice agent mode to Codex

OpenAI is introducing a full voice agent mode for Codex, leveraging the capabilities of the new GPT Live model. This update allows developers to engage in real-time conversational audio interactions with the coding assistant, enabling hands-free coding, debugging, and planning workflows.

// ANALYSIS

Bringing real-time voice interaction to AI coding agents could significantly speed up conceptual planning and debugging, though precise syntax handling via audio remains an operational challenge.

–Powers real-time voice interaction using OpenAI's GPT Live model.
–Enables developers to query, debug, and architect code using conversational audio commands.
–Signals OpenAI's strategy to expand multimodal agent capabilities across developer tooling.

// TAGS

openaicodexgpt-livevoice-agentaidevtool

DISCOVERED

2h ago

2026-07-23

PUBLISHED

3h ago

2026-07-23

RELEVANCE

8/ 10

AUTHOR

mark_k

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

UPDATE22m ago

Qoder offers 14-day Pro trial with 300 credits

Qoder, Alibaba's agentic coding platform available across desktop IDE, CLI, JetBrains plugin, and cloud environments, is offering new users a 14-day Pro trial with 300 credits. The trial provides discounted access to models like Qwen 3.8 Max, enabling developers to handle complex multi-file code editing and task delegation.

UPDATE22m ago

Qoder slashes Qwen 3.8 Max credit costs 98%

Qoder is offering a major promotional discount on Alibaba's 2.4-trillion-parameter foundation model, Qwen 3.8 Max, lowering its credit billing coefficient to 0.05x during standard hours and 0.01x during off-peak hours. Running over a 14-day period, this campaign allows developers to leverage high-capacity coding and reasoning intelligence within the Qoder agentic IDE at a fraction of standard operational costs.

INFRA41m ago

SGLang accelerates LLM serving with RadixAttention

SGLang is an open-source inference framework designed to accelerate and optimize large language model serving at scale. By combining a flexible programming frontend with a high-performance runtime featuring RadixAttention, SGLang enables complex agentic workflows while lowering latency.