TECHNOLOGY PROFILE · TECH-0148
llama.cpp
Portable C/C++ inference engine for supported GGUF models.
Core capability checked in primary documentation. This profile preserves the prepared catalog’s original descriptions. Advantages depend on the workload; comparisons are editorial, not benchmark results.
Where it has an edge
Quantized CPU/GPU hybrid execution can run local models without a full Python training stack.
Best-fit problems
Offline language models and machines with constrained GPU memory.
Tradeoffs & cautions
Serving throughput and model coverage differ from dedicated GPU servers.
What to test before choosing
Test the actual target device offline; measure load time, memory, battery/thermal behavior and task accuracy.
Runtime & hardware
Local device; supported CPU/GPU/NPU
Access & licensing
Mixed licenses; check model and app terms
Execution
Local / self-hosted
Lifecycle
Not independently audited
Comparison basis
Editorial problem-fit judgement; no head-to-head benchmark performed
Connected concepts
Alternatives to investigate
Ollama · vLLM · MLX