Model Releases
b11958: server : support tool calls in /v1/realtime (#31287)
**Issue Summary (b11958 – Server Support for Tool Calls in /v1/realtime)** The issue requests adding server‑side handling of OpenAI API “tool call” functionality to the `GET/POST /v1/realtime` endpoin
Issue Summary (b11958 – Server Support for Tool Calls in /v1/realtime)
The issue requests adding server‑side handling of OpenAI API “tool call” functionality to the GET/POST /v1/realtime endpoint, allowing language models to return structured tool‑calling payloads that can be executed by downstream systems. It requires updating the real‑time server logic, ensuring compatibility with existing clients and documentation, and integrating error handling for unsupported or misformatted calls. The goal is to enable developers to use real‑time LLM outputs as actionable commands directly within their applications.
Related
Source: llama.cpp Releases | 2026-10-28