# llama.cpp v0.?? removes deprecated APIs and promotes speculative NextN feature

> The open‑source llama.cpp library drops old functions and moves the NextN speculative generation API into the main header.

Oossa · 2026-11-13 · https://oossa.com/en/llama-cpp-v0-removes-deprecated-apis-and-promotes-speculative-nextn-feature

The ggml‑org/llama.cpp project released a new version on 2026-11-13. The update deletes several deprecated APIs, including llama_set_warmup and the defrag_thold option. It also moves the speculative NextN API from the auxiliary llama‑ext.h file into the primary llama.h header and renames related functions. Documentation for the llama_model_dflash_selector_top_k function was added.

## The facts

- Release date: 2026-11-13 ("Fri Nov 13 2026 10:43:20 GMT+0100")
- Deprecated APIs removed: llama_set_warmup, defrag_thold, and related header includes

## Why it matters

Developers can now use the NextN speculative generation directly from the main API, simplifying code and reducing reliance on outdated functions.

## Sources & references

1. [ggml-org/llama.cpp b12347](https://github.com/ggml-org/llama.cpp/releases/tag/b12347) – llama.cpp, 2026-11-13

Last updated: 2026-11-13