Part three measures Qwen3.8 Flash Next story generation, sustained throughput, and patched long-prompt performance on an Apple M4 Max.
Topic
ollama
Part two compares local LLM input, output, latency, memory, and sustained performance on an Apple M4 Max, with dated Claude and GPT-5.6 Sol API speed references.