InData Science CollectivebyAnkit Agrawal·4d ago5 Ollama and MLX Settings to Optimise for a Local LLM on a Mac in 2026I measured what each Ollama and llama.cpp setting costs in memory on a 24GB M5 MacBook Pro. Most are set for you before you type a prompt.A response icon1A response icon1
InSalt IndexbyRothyyorn·Oct 29Ollama vs MLX: I Compared Speed, Memory Use, and Setup on MacCompare Ollama and MLX on Apple Silicon to understand local LLM speed, memory consumption, setup complexity, and which workflow makes the…A response icon1A response icon1
Paul Hackenberger·Oct 29🕵️ A fast, reliable, offline Community Moderator Agent — Part IV: Kontext is KeyPaul Hackenberger · 28.10.2026 · 20 min readA response icon1A response icon1
Paul Hackenberger·Oct 22🕵️ A fast, reliable, offline Community Moderator Agent — Part III: FinetuningPaul Hackenberger · 22.10.2026 · 10 min read
Varun Nuthalapati·Oct 21Ollama Now Speaks “Decision”: Jev-Style Models Arrive in the CLI, on MLX by Defaultollama run — ask puts Kev, Nimble, and Clef one command away on a Mac, and the way each model reads your question matters more than its…
InData Science CollectivebyAnkit Agrawal·Oct 19Can you run Meta’s Muse Spark Open on a Mac? How much RAM 816B really needsI sized every GGUF and MLX build against every Mac. A 256GB Studio loads a 1-bit build; 512GB is where it gets real.A response icon1A response icon1
InMateeDevsbyLukas Zima·Oct 14Building an LLM from scratch in MLX SwiftWith all the theory behind itA response icon1A response icon1
Nandana Dileep·Oct 1113 Performance Experiments on One Reranker: The Limits of Faster GPU CodeBenchmarking compilation, custom kernels, batching, and serving
