Someone Fine-Tuned OpenBMB’s MiniCPM5-1B on Claude Fable 5 Traces to Ship a 657MB Local Thinking Model

A community developer released MiniCPM5-1B-Claude-Opus-Fable5-Thinking, a 1B-parameter fine-tune of MiniCPM5-1B.

The model is packaged as GGUF for local use and works with runtimes like llama.cpp, Ollama, LM Studio, jan, and KoboldCpp.

It was trained from generated Claude Fable 5 conversation traces, aiming for reasoning-style behavior without cloud access or an API key.

Multiple quantizations are available, including a 657MB Q4_K_M build and support for up to 128K context.