# \[ANN\] Haskell bindings for llama.cpp — llama-cpp-hs

**URL:** <https://discourse.haskell.org/t/ann-haskell-bindings-for-llama-cpp-llama-cpp-hs/12136>\
**Category:** Announcements\
**Created:** [May 21, 2025, 9:55am UTC](https://discourse.haskell.org/t/ann-haskell-bindings-for-llama-cpp-llama-cpp-hs/12136 "2025-05-21T09:55:06Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![tusharad](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.haskell.org/tusharad/32/5133_2.png) [@tusharad](https://discourse.haskell.org/u/tusharad)\
**Post date:** [May 21, 2025, 9:55am UTC](https://discourse.haskell.org/t/ann-haskell-bindings-for-llama-cpp-llama-cpp-hs/12136/1 "2025-05-21T09:55:06Z")

</div>

Hey folks,

I’m excited to share the initial release of [llama-cpp-hs](https://github.com/tusharad/llama-cpp-hs) — low-level Haskell FFI bindings to [llama.cpp](https://github.com/ggml-org/llama.cpp), the blazing-fast inference library for running LLaMA and other local LLMs.

**What it is:**

- Thin, direct bindings to the `llama.cpp` C API
- Early stage and still evolving
- Most FFI is "vibe-coded"™ — I’m gradually refining, testing, and wrapping things properly
- That said, basic inference examples are already working!

🔗 [GitHub](https://github.com/tusharad/llama-cpp-hs)  
📦 [Hackage](https://hackage.haskell.org/package/llama-cpp-hs)

Contributions, testing, and feedback welcome!