Thinking Machines added Moonshot AI's Kimi K2.5 to Tinker with text input supported initially. The underlying Tinker announcement, quoted by the company, told users to update the software development kit and described vision support as forthcoming.
This made Kimi K2.5 another model available through the training service, but the announced integration had a narrower input scope than a multimodalA model that works with more than text — reading images, audio, or video, and sometimes generating them too.Full definition → workflow: experiments could use text, while image input remained a planned extension.
Terms in this piece · Glossary
multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
fine-tuning — Taking a trained model and training it a bit more on your own examples so it gets better at one specific job.
