Best open-source AI models, by use case

Curated lists for the six workloads we get asked about most. Each page explains what to optimise for, names the 3–6 strongest current picks, and flags the gotchas you'll otherwise discover the hard way.

Best for

Coding

Open-source models that excel at code generation, completion, and review. Picks span on-device 7B models through frontier-class 30B+ specialists.

Best for

Local inference

Open-source models that run well on consumer hardware (RTX 4090, Apple Silicon, even laptop iGPUs). Picks balance quality with VRAM footprint.

Best for

RAG

Models tuned for retrieval-augmented generation: long context, strong instruction following, native citation behaviour where possible.

Best for

Vision

Open-weights multimodal models that can read images, do OCR-adjacent tasks, and reason over diagrams and screenshots.

Best for

Reasoning

Models that generate long internal chains-of-thought before answering — designed for math, code, and multi-step problem solving.

Best for

Edge

Sub-4B models that run on phones, laptops, and embedded devices. Optimised for memory footprint and tokens-per-second on integrated GPUs.