SILICON VALLEY -- November 13, 2026 -- Meshy.ai's Make Printable feature cut the share of 3D prints with major missing features or holes from 44% (11 of 25) to 8% (2 of 25) in a blinded FDM printing test run by the University of Massachusetts Amherst's Advanced Digital Design and Fabrication facility (ADDFab), the company said. That is 9 fewer failed prints, a drop of about 82% by our calculation, measured on prompts Meshy.ai selected to be hard to print.
Meshy.ai commissioned and funded the study and supplied six Bambu Lab A1 printers and consumables. The release states that ADDFab's participation is not an endorsement of Meshy.ai products or an audit of its software.
Test printed 125 parts across five workflows with identities hidden
Meshy.ai generated models for 25 prompts, and ADDFab sliced and printed 125 parts. Each prompt ran through five workflows: Meshy 7, Meshy 7 with Make Printable, and three other AI 3D generators. Settings and filament were identical for every workflow, and engine identities were concealed during file preparation, printing and grading.
Three experienced 3D printing users scored each part against the lab's five-criterion rubric. A further 17 volunteers ranked the anonymous prints. ADDFab Director Dave Follette said the aim was to test, by printing on a consumer FDM printer, whether the platforms can produce parts that are both visually convincing and manufacturable.
Make Printable lifted Meshy 7 from the bottom of the field to the top
Raw Meshy 7 produced 14 of 25 prints (56%) without major missing features or holes. That matches the weakest of the three other generators, which also scored 14 of 25. With Make Printable, Meshy 7 reached 23 of 25 (92%).
The three other generators scored 21 of 25 (84%), 15 of 25 (60%) and 14 of 25 (56%). Meshy 7 with Make Printable therefore led them by 8, 32 and 36 percentage points. In print counts, rivals had 4, 10 and 11 prints with major defects against 2. With 25 prompts per workflow, each print equals 4 percentage points, so the lead over the closest rival is two prints.
Two Make Printable prints still failed: a six-legged insect and a long-finned tropical fish. Meshy.ai says other defects may remain in the prints counted as passing.
Software check shows complete surfaces rising from 33.4% to 98.8%
Meshy.ai also ran a software assessment of model geometry across 100 prompts. Models with complete, connected surfaces rose from 33.4% for raw Meshy 7 output to 98.8% after Make Printable, an increase of 65.4 percentage points, or roughly three times the share. The check counts detected openings and invalid connections. It does not measure physical print success.
Make Printable repairs mesh defects and reinforces thin features for the selected size and process. It works on models generated in Meshy.ai and on uploaded models from other tools. Developers can reach it through the Repair Printability API, which accepts external model files. The company says material, orientation, support strategy, printer calibration and settings still affect print outcomes. Make Printable is available now as part of Print Check in Meshy.ai.
Prompts were chosen to defeat rivals, and only FDM was tested
Meshy.ai and ADDFab selected the 25 prompts to challenge delicate-feature reproduction, prioritizing prompts for which competing models were predicted to be too thin to print. The study says the findings should not be read as representative of all prompts. Meshy.ai did not name the three competing generators.
The physical test covered FDM printing only, although Make Printable also offers SLA and Full-Color presets. Future testing could extend to SLA, SLS and Full-Color printing, according to the study.