Inkling 2 Small Prompting: Set Thinking Effort First

Thinking Machines has published the Inkling 2 Small weights on Hugging Face under Apache 2.0, after launching the model on Tinker on Nov 5. It is a 594B-parameter Mixture-of-Experts model with 17B active per token, text and image input, and a context window of up to 1M tokens. The launch announcement says it supports variable effort, and the lab ran its own evaluations at effort 0.99 and temperature 1.0. Thinking Machines' Inkling cookbook for Tinker says to set thinking effort (a number from 0 up to, but not including, 1; 0.9 by default) through the renderer, not by typing the effort message into the prompt. Set the effort in the API call, not the prompt. The Hugging Face model card does not mention effort, so confirm the setting name in the docs for whatever you serve the model with, and spend the prompt on the goal and the output format.

Review the diff in <diff>{{DIFF}}</diff> for bugs that a unit test would miss. For each finding, give the file and line, a one-sentence cause, and a one-line fix. If there are no findings, reply with exactly: No findings. Do not restate the diff or list style nits.