Meshy’s 3D Agent Remains in Public Beta
Meshy’s conversational 3D Agent is currently available in public beta for Pro members and above, combining concept generation, modeling and export in one workflow.
From prompt to production asset
Meshy’s 3D Agent remains in public beta for Meshy Pro members and above. The product is aimed at a workflow that has traditionally required several separate tools: developing an idea, generating visual references, producing a 3D asset, refining it and exporting it for printing or game development.
The agent accepts natural-language instructions as well as photos and sketches. It can propose creative directions, generate batches of concepts, turn selected designs into models and continue refining them through conversation. Meshy says users can export results in formats including FBX, OBJ, GLB, USDZ, STL, BLEND, 3MF and DXF, covering game engines, digital-content tools and 3D-printing workflows.
Why this is more than another text-to-3D feature
The significant change is the product boundary. Earlier generative 3D tools typically stopped after producing a single asset. Meshy’s agent is designed to coordinate several steps and preserve context between them. That makes it closer to a specialized creative assistant than a prompt box attached to a model.
The company’s own product materials describe use cases ranging from indie games and product concepts to printable objects and batch asset creation. The same workflow can also isolate multiple objects from a reference image and prepare them as separate models, reducing some of the manual setup work that slows down small creative teams.
The open question
Meshy’s public materials say a broader free trial is planned, but the product is not yet generally available. The current beta will test whether conversational orchestration actually improves production speed rather than simply making demos easier. Quality consistency, topology, material accuracy and licensing remain important once assets enter commercial pipelines. Meshy’s strongest claim is therefore not that it can generate a model, but that it can connect ideation to a usable deliverable. The market will judge that claim in finished projects, not isolated previews.