Thinking Machines: Inkling Small (free)

Thinking Machines: Inkling Small (free) is a text model from Thinking Machines. It is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. The provider describes it as the smaller, more efficient member of its family, which tells you where it sits in the lineup: a compact option rather than the largest model the lab offers.

The model accepts text, image and audio as input, and it produces text only. That combination makes it a multimodal reader rather than a generator of media: you can hand it written prompts, pictures or audio and get written answers back. Its context window is 1,048,576 tokens, and the maximum output is 262,144 tokens. The context length is the headline number here, since it is large enough to hold very long documents, long transcripts or extended conversations in a single request without chunking.

Those capabilities point to a fairly specific set of uses. Long document work is the obvious one: summarising, extracting structured data from, or answering questions over large files that would otherwise need to be split across many calls. Because images and audio are accepted as input, the same model can handle tasks like describing or extracting text from images, or working over recorded speech, without a separate transcription step. Reasoning support means it can also be used for problems where you want the model to work through steps before answering, and tool calling lets it sit inside an agent loop that reaches out to external functions.

You call it through api14's OpenAI-compatible API using the model id "thinkingmachines/inkling-small:free". The supported request parameters are frequency_penalty, include_reasoning, max_tokens, presence_penalty, reasoning, reasoning_effort, seed, stop, temperature, tools and top_p. In practice that covers the usual sampling controls (temperature, top_p, frequency and presence penalties, stop sequences, seed for reproducibility), output length via max_tokens, and the reasoning and tool-related options: reasoning and include_reasoning to control whether reasoning is returned, reasoning_effort to steer how much effort is spent, and tools for function calling.

Two things are worth checking before you commit. The model outputs text only, so if you need generated images or audio, it is not the right fit. And while the context window is very large, the free variant carries the ":free" suffix in its model id, which is a signal about how it is offered rather than a statement about capability; if you need a larger or more capable member of the same family, compare against the other Inkling models on api14.