DLLM - D Language 🤖 on 🦙.cpp

Danny Arends Danny.Arends at gmail.com
Wed Mar 25 20:18:28 UTC 2026


On Wednesday, 25 March 2026 at 12:16:08 UTC, Sergey wrote:
> On Tuesday, 24 March 2026 at 21:02:34 UTC, Danny Arends wrote:
>> However, the current version can read and "understand" it's 
>> own codebase with just an 8B model. So you could put it in a 
>> vibe
>
> Nice
>
> Btw there is another vibe-coded bindings recently released
> https://code.dlang.org/packages/llama-cpp-d

Interesting, but those are just bindings. But, binding llama.cpp 
doesn't make a coding agent. DLLM provides:

- local RAG with binary-persisted embeddings and cosine 
similarity ranking
- Vision support via mtmd (multimodal - load an image, ask about 
it)
- Docker-sandboxed code execution (Python, JavaScript, Bash, R, D)
- Web search via SearxNG, as well as web fetch ability
- File I/O, date/time, encoding, audio playback tools
- KV cache condensation via a dedicated summary model
- Thinking budget enforcement via token limits
- Memento system - the agent writes notes to its future self 
between sessions
- Full interactive and oneshot modes

And again, not vibe coded 😉. It's a tool for vibe coding.

See my blog post: 
https://dannyarends.nl/DLLM%20-%20D%20Language%20%F0%9F%A4%96%20on%20%F0%9F%A6%99.cpp




More information about the Digitalmars-d-announce mailing list