Open Models Almanac
Methodsinference & servingagentic scaffolding

specific method · filed under inference & serving

Python tool for visual tasks

A Python code interface provides image operations such as cropping, zooming, and visual probing to support image understanding.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See agentic scaffolding for the methods filed beside it.

Also called Python Tool.

sources
3
models
2
labs adopt it
2
strongest
used

Documented in

Thinking Machines Lab

Evidence

3 quotes from 3 sources, one card per model, each opening with its strongest.

How sources treat it
optional 1used 2

Step-3.7-Flash

StepFunadopted2 sources

we grant the model an enriched action space to interact with images, including cropping, zooming in and out, and drawing pixels or bounding boxes. These tools are implemented as a unified code interface, commonly referred to in the field as the Python tool.

usedinference servingin Step 3.7 FlashStepFun
Step 3.7 Flash·StepFun·2026-05-09
1 more from this model

Python Tool — For fine-grained visual tasks (high-resolution images, visual probing, bounding-box analysis)

optionalinference servingin Step 3.7 FlashStepFun

Inkling

Thinking Machines Labadopted1 source

During inference, Inkling can also leverage a Python tool to support image understanding through operations such as zooming and cropping, while seamlessly integrating visual reasoning with code-based reasoning.

usedinference servingin InklingThinking Machines Lab
↑