Molmo models combine visual input and question. In supported versions, you can request the coordinates of the object, and in Molmo 2 you can also work with video.
A question that can be answered by a point
A warehouse photo shows two similar boxes. Calling both brown is of little use when the operator needs the one on the upper shelf. Molmo can process the image and question together, and its pointing mode returns a location. Overlay the point on the original frame to check it. If your application resizes or crops the image, verify the coordinate conversion.
Two items and an empty frame
Photograph a shelf with two boxes, then remove the upper one and take another photo. For the first frame, specify which location to look at. Repeat the same question for the second frame. Check separately whether the model points to the right box and how it responds when that box is absent. Add a photo where one object partly blocks another. A point on the visible part does not indicate the area of the hidden box.
How to Expand Verification
For multiple frames, pick a Molmo variant with video support documented by Ai2 and verify the timestamps. Record the original size, query wording, and weights. Treat scene answers, coordinates, and exact outlines as different outputs. Use a separate method for outlines. Test different lighting and viewpoints. Count misses separately from incorrect points.
Suitable tasks
- Check the indication of the desired object among similar
- Verify answers to questions about content
Strengths
- The model can return an indication of the object's location
- Ai2 publishes weights and launch examples
Limitations
- The point does not specify the exact contour of the object
- Occlusion and similar items require separate checks
Selection and conditions
- Free access
- Yes
- Trial period
- Not confirmed
- No sign-up required
- Not confirmed
- Russian interface
- Not confirmed
- Local installation
- Yes
- API
- Not confirmed
“Not confirmed” means there is not enough information. The feature may still be available; check the developer’s terms.
- Input formats
- Image, Text
- Result
- Data, Text
Terms
The conditions of the available weights are indicated in the card of the corresponding model. Their equipment or managed calculations are paid separately. The point does not specify the exact contour of the object.From task to first result
Practical example
- Task
- Check the indication of the desired object among similar
- What to prepare
- His photo with two similar boxes, one of which stands on the top shelf.
Task brief or prompt
Point the box on the top shelf. If the object is not visible, do not transfer coordinates from the other box.
Expected result
Coordinates that can be applied to the source frame for verification.
How to check the result
- Verify coordinate conversion after resizing the image.
- Check the second frame without the right box.
An editorial teaching example of a task brief, not the result of hands-on testing of the service. Check that the required mode is available, that you have rights to the source material and that the intended use of the result is permitted.
Check details
Official sources
This description is based on the developer’s material. Features, prices and availability may change; check the primary source before use.