Tmappdev releases Lang-Segment-Anything for prompt-based image segmentation

On October 3, 2026, DEV Community's AI section reported that Tmappdev released a model called Lang-Segment-Anything on Replicate. According to the report, the model combines natural-language prompts with image segmentation: rather than supplying spatial coordinates, a user describes the target in words and the model uses that description to identify and isolate the object in an image.

The report says Lang-Segment-Anything is built on Segment Anything while adding natural-language understanding, so that language prompts and segmentation work together instead of being handled as separate steps. Tmappdev is named as the maintainer of the model.

The write-up is presented as a simplified guide to the model. In it, the author points readers to AImodels.fyi and invites them to follow the account on Twitter for similar analyses of AI models.

Beyond those points, the material available does not state a version number, parameter count, training data, benchmark results, or pricing for Lang-Segment-Anything, and no other source in the material reports on the release. What the record shows so far is simply that the model was published on Replicate, as described in that single October 3 report.