Context-scaling benchmark of Qwen 3.8 27B in NVFP4/MLX through Ollama on a 36 GB M4 Max for agentic coding, measuring TTFT, prefill and decode from 4K to 115K.
Start with your machine.
Capacity is the reported total for this setup. Multi-GPU and unified-memory layouts can behave differently.
A daily local coder on an RTX 5060 8GB and 32GB of RAM finds responsiveness under memory pressure matters more than benchmark speed.
A 3090 owner runs Qwen 3.8 27B daily in Cline at a 96K context by quantising the KV cache to q8_0, reporting about 43 tok/s once the model stops thinking.
A small GPU runs sub-10B models through Ollama, with a custom skill and memory layer driving a broad set of offline personal and automation workflows.
A model-agnostic personal automation system switched from a cloud model to Qwen 3.5 9B on an M1 Pro MacBook with 16GB of unified memory, handled a real subset of agent work locally, and ran 0.8B/2B models on an iPhone.