dynalm-project

dynalm-project/dynalm

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

3 stars 2 forks 3 watchers C++ Apache License 2.0
apple-silicon continuous-batching cpu-inference gemma gguf inference-engine llama llama-cpp llm llm-inference local-ai local-llm metal mixture-of-experts ollama openai-api quantization qwen self-hosted vulkan
View on GitHub Website
13 Open Issues Need Help Last updated: Oct 22, 2026

Open Issues Need Help

View All on GitHub
dynalm pull: show progress while checking the sha256 8 days ago
enhancement good first issue
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Structured output: support minItems and maxItems in JSON Schema 8 days ago
enhancement good first issue
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Docs: use DynaLM with Open WebUI 8 days ago
documentation good first issue
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Call for testers: Linux, Windows and Intel Macs 8 days ago
help wanted testing platform
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Share dynalm benchmark results from your hardware 8 days ago
good first issue help wanted benchmarks
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
NEON kernels for Q2_K, Q3_K and Q5_K (DynaCore exact mode) 9 days ago
enhancement help wanted
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Model support: OLMo 2 14 days ago
help wanted models
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
`dynalm pull`: download split (sharded) GGUF files 14 days ago
enhancement help wanted models
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Shorten long header comments 14 days ago
good first issue
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Docs: guide for using DynaLM with Open WebUI 14 days ago
documentation good first issue
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Share `dynalm benchmark` numbers from your hardware 14 days ago
good first issue benchmarks
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Test the Vulkan build on a real AMD, Intel or NVIDIA GPU 14 days ago
help wanted testing gpu
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan
Call for testers: DynaLM 0.2.0 on Linux, Windows and Intel Macs 14 days ago
help wanted testing platform
dynalm-project/dynalm
3

Run LLMs locally on Mac, Linux and Windows: an open-source llama.cpp and Ollama alternative with Metal GPU support, an OpenAI-compatible API server and continuous batching. Llama, Qwen, Gemma, Phi, Mistral, Granite (GGUF).

C++
#apple-silicon#continuous-batching#cpu-inference#gemma#gguf#inference-engine#llama#llama-cpp#llm#llm-inference#local-ai#local-llm#metal#mixture-of-experts#ollama#openai-api#quantization#qwen#self-hosted#vulkan