Text-to-3D model generation with texturing for game devs.
3D & AR
Meshy is the text-to-3D and image-to-3D generation platform that spent 2023-2026 methodically outbuilding every other AI 3D tool on the axis game studios actually pay for — the pipeline. Where Luma Genie treats generative 3D as a fast, light, single-object generator and Spline AI treats it as one feature inside a web design canvas, Meshy has been the tool trying hardest to answer the practical question a technical artist asks on day one: "Can I actually put this asset in my Unreal project without three days of cleanup?"
The 2026 answer is a qualified yes. Meshy's current model produces meshes with reasonable topology, PBR channel outputs (base color, roughness, metallic, normal), an automatic retopology pass on Pro tiers, and — since the 2025 Meshy 5 release — automatic rigging on humanoids and quadrupeds that plugs into Mixamo-style animation libraries with one export. None of that is production-final on a AAA title. Most of it is usable on an indie release, a game jam, an XR prototype, or an e-commerce 3D catalog. That is a different bar than Luma Genie's "concept art in ten seconds" bar, and it is a bar that pays.
Founded by Ethan Hu in 2023 and quietly stacking releases through the Meshy 3, 4, and 5 model generations, the company has settled into a lane the frontier labs are not competing in: mid-tier generative 3D specifically for game and interactive workflows. That focus shows up in every corner of the product — from the retopology dial to the animation preview to the two-step image-to-3D-with-refinement flow — and it is why Meshy is now the default recommendation for anyone building a game, an AR filter, or a virtual product visualization who does not have modeling budget.
The one-line positioning: Meshy is the text-to-3D and image-to-3D generator for creators whose endpoint is an interactive engine, not a mood board. If your output is Unity, Unreal, Godot, Blender, or a WebGL scene, Meshy is where you should be looking first in 2026.
Meshy's product surface is unusually complete for a category this young. Feature depth is where the real differentiation sits.
Text-to-3D generation. The core flow: type a prompt, pick between four initial mesh candidates in around 30-60 seconds, then choose one and run a refinement pass that produces a higher-quality mesh with PBR textures over another 2-3 minutes. Two-stage generation is a design choice that has aged well — the preview step is fast and cheap, the refinement step is where the quality shows up.
Image-to-3D generation. Upload a reference image (or four reference views for multi-view mode) and Meshy reconstructs a mesh approximating the subject. Multi-view mode with front/back/left/right references is the setup a concept artist should use — quality gap between single-view and multi-view is enormous. This is where Meshy meaningfully outperforms Luma Genie in 2026.
PBR texture channel output. Meshes ship with proper PBR maps: base color, roughness, metallic, and normal. Drop the mesh into Unity's Universal Render Pipeline or Unreal's default material graph and the surfacing behaves the way an artist would expect. This is the single feature that most separates Meshy from Genie's baked-diffuse output, and it is why studios pick Meshy for real work.
Automatic retopology. On Pro tiers, Meshy runs a retopology pass that converts the generative mesh into cleaner quads with more even edge flow. Not production-final for hero characters, but good enough to skip a manual retopo step for background props, environment fill, and second-tier assets.
Automatic rigging and animation preview. Meshy 5 (2025) shipped automatic rigging for humanoids and quadrupeds. Generate a character, run auto-rig, preview a walk cycle in the browser, and export to a rigged FBX ready for Mixamo or a custom animation library. The rig quality is not what a technical animator would hand-place, but the "from prompt to walking character in seven minutes" pipeline is a real product.
Multi-format export. GLB, FBX, OBJ, USDZ, STL, and Blender-friendly formats. The FBX support alone is a reason Unity teams still on FBX workflows pick Meshy over Genie.
Style controls. Realistic, cartoon, sculpt, and voxel style presets, each with meaningful visual differences. The voxel preset in particular is popular with Minecraft-style indie devs and produces geometry that actually reads as low-poly rather than generative-mesh-approximating-low-poly.
Public library and community remix. Community-generated meshes are browsable and CC-licensed within the platform. Useful for warming up on prompt patterns before you spend credits on your own generations, and occasionally a source of assets you can license commercially.
API access. Available on Business tiers and above. Real integration surface — a small studio can wire Meshy generation into a batch pipeline for e-commerce 3D catalogs or bulk asset creation.
Discord community. Active Discord with prompt sharing, technique threads, and quick community feedback on outputs. Real user community, not a growth-hack forum.
Meshy publishes four tiers plus custom Enterprise. Pricing is monthly with a meaningful annual discount.
| Plan | Price (annual, per month) | Best for |
|---|---|---|
| Free | $0 | Trying it out — 200 credits/month, community license |
| Pro | $20 | Solo creators — ~1,000 credits/month, commercial license |
| Max | $60 | Regular indie devs — ~4,000 credits/month, private generations, faster queue |
| Ultimate | $120 | Small studios — ~10,000 credits/month, API access |
| Enterprise | Custom | Studios with volume needs, custom SLAs, dedicated support |
Meshy prices on credits, and a typical text-to-3D generation costs about 5 credits (preview) + 20 credits (refined) — so a Pro plan lands around 40 finished assets per month. Image-to-3D and multi-view flows cost slightly more; retopology and rigging cost a few credits each. The math scales linearly and predictably, which is more than can be said for most credit-based AI billing in 2026.
The Free tier at 200 credits is enough to genuinely evaluate the product — you can produce five or six refined assets before you have to decide about a subscription. Pro at $20/month is the standard indie tier and where most solo game devs land. Max at $60 is the right choice the moment your monthly output exceeds ten refined assets — the per-credit price improves substantially and the private-generation flag matters if you are shipping proprietary work. Ultimate at $120 is where the API tier unlocks; below that, you are clicking generations in the web UI.
Commercial licensing is included on all paid tiers. The Free tier is Creative Commons — fine for personal projects and portfolio pieces, not fine for a commercial game release.
Compared to Luma Genie's shared-credit-with-Dream-Machine model (roughly $30/month for equivalent 3D volume), Meshy Pro at $20 is a better deal if you only care about 3D. Compared to Kaedim's studio pricing (several hundred dollars a month for human-in-the-loop cleanup), Meshy Ultimate at $120 is dramatically cheaper for teams doing pure-AI generation at volume.
Pros
Cons
Indie game developers on a modeling budget. Meshy is the tool a two-person indie team uses to fill an environment with rocks, crates, foliage variants, and background props without hiring a modeler. The refined-mesh + PBR + retopo combination genuinely ships in released games.
Game jam and prototype teams. For a 48-hour jam, Meshy can populate an entire game world with themed props in a Saturday morning. The GLB and FBX exports drop into Unity, Godot, and Bevy pipelines without conversion.
AR filter and Snapchat Lens creators. USDZ export straight into Apple's Reality Composer or Snapchat's Lens Studio, no format conversion. Rigged characters export into humanoid AR templates fast.
E-commerce 3D catalog teams. Image-to-3D with the API on Ultimate is the pipeline behind a growing number of small e-commerce sites building 3D product previews from existing product photography. Real revenue-side ROI.
Concept artists producing 3D turntables. For concept artists whose deliverable used to be flat 2D art and is now a rotating 3D preview, Meshy is the fastest path from sketch to shareable turntable.
Educators teaching 3D pipelines. The Meshy → Blender retopo → Substance texture workflow is a full pipeline in miniature, cheap enough for a classroom and complete enough to actually teach the concepts.
Studios needing pre-viz assets fast. Even studios that ship hand-modeled hero assets use Meshy for pre-viz — mock up the scene composition, prove the shot works, then replace with real assets. Genuinely useful upstream tool.
Do not use Meshy for: AAA hero characters (still a human sculpt job), high-frequency scene composition (Spline AI is better for interactive web 3D), or ultra-fast mood-board work where Luma Genie has a UX advantage.
Luma Genie — the "start here" alternative. Genie is faster to feel productive with, has a unified account with Dream Machine video, and prices well for creators using both. Weaker on PBR, retopo, and rigging — which is the whole reason Meshy exists. Use Genie for mood boards, Meshy for finals.
Spline AI — different product. Spline is a web-native 3D scene composer with AI features; Meshy is a single-object generator. If you are building an interactive 3D webpage, Spline is the answer. If you are producing 3D assets to drop into an engine, Meshy is the answer.
Kaedim — the human-in-the-loop 3D generation service. Priced for studios ($300-$1000+/month), higher-quality retopology because a real technical artist reviews every mesh, longer turnaround. Different product entirely; the honest comparison if quality matters more than speed.
Rodin (Deemos) — a Chinese-founded competitor with strong hard-surface and character output in 2026. Meshy is more polished as a product; Rodin sometimes produces sharper results on specific subjects.
Alpha3D — enterprise-focused for e-commerce and product visualization. Different buyer.
Sloyd — parametric procedural 3D generation, not AI-generated in the same sense. Useful complement for procedural game assets (buildings, weapons, vehicles) where consistency matters more than uniqueness.
Stable 3D / Stable Zero123 — open-source-adjacent generative 3D from Stability AI. Rougher output, more control if you self-host, useful for research pipelines.
For most AIQORA readers building a real game, filter, or interactive 3D product: try Meshy Free tier first, then commit to Pro or Max based on your monthly output. If your goal is concept-only, Luma Genie is a lighter tool and probably cheaper for your volume.
Sign up at meshy.ai with Google or email. Complete the onboarding — it walks through text-to-3D, image-to-3D, and the refinement flow using free credits, which is exactly the right way to learn what the product does before you commit.
Start with a text-to-3D prompt from a category you actually care about — for game devs, try "a stylized wooden crate, weathered oak, low-poly, PBR" or "a hand-forged iron dagger, brass pommel, leather-wrapped grip." Pick the best of the four preview candidates, then run the refinement pass. When the refined mesh finishes, download the GLB and open it in Blender to inspect topology from a wireframe view. That inspection habit is the fastest way to learn what to expect from the model.
Try image-to-3D with a multi-view setup next — four reference images from front, back, left, right. Quality gap between single-view and multi-view is the single biggest lever you have on output quality; make it a habit.
If you plan to ship anything commercially, upgrade to Pro before you generate anything you want to use. The commercial license kicks in the moment you subscribe; the Free tier's CC license is fine for personal use but not for a released product.
A practical tip: prompt with material and style specificity. "Stone gargoyle" produces something forgettable. "Weathered granite gargoyle, gothic cathedral style, PBR realistic, subtle moss growth on stone" produces something you can drop into a scene. Meshy responds well to material and lighting specificity in the prompt.
Is Meshy free? Yes — the Free tier gives you 200 credits per month and CC-licensed output. Real work with commercial licensing happens on Pro at $20/month annual.
Can I use Meshy output commercially? On paid plans, yes, with full commercial rights across game, film, e-commerce, and AR use. Free tier is CC — personal and portfolio use only.
Does Meshy export PBR textures? Yes — base color, roughness, metallic, and normal maps ship with refined meshes. This is Meshy's biggest differentiator against Luma Genie.
Does Meshy do automatic rigging? Yes — Meshy 5 (2025) added humanoid and quadruped auto-rigging. Rigged output exports to FBX ready for Mixamo animations.
What formats does Meshy export? GLB, FBX, OBJ, USDZ, STL, and Blender-friendly formats. FBX support is a Unity-team win.
Is there an API? Yes, on Ultimate ($120/month) and Enterprise tiers. Real API for batch pipelines and integrations.
How does Meshy compare to Luma Genie? Meshy has PBR, retopo, and rigging; Genie has a lighter UX and unified billing with Dream Machine video. Genie is a better mood-board tool; Meshy is a better asset factory.
Is the mesh topology production-ready? For secondary assets, yes. For AAA hero characters, no — you will still want a human retopo pass. This is the right expectation to have across every text-to-3D tool in 2026.
Meshy on the Pro plan at $20/month is the sharpest text-to-3D tool for creators whose endpoint is an interactive engine. The PBR channel output alone earns its price; the automatic retopology, the auto-rigging, and the multi-view image-to-3D compound on top of that to make Meshy the most complete pipeline product in the generative 3D category as of 2026.
For indie game devs, Meshy is the tool that finally makes "I do not have modeling budget" a solvable problem instead of an existential one. For AR creators, it is the fastest path from concept to a rigged character in Reality Composer. For e-commerce teams building 3D product catalogs, the API on Ultimate is the difference between a manual process and a pipeline.
Where Meshy stops being the sharpest answer: hero-character work on AAA titles still needs a human sculptor; scene composition for interactive web is better handled by Spline AI; and the fastest "I just want a mood-board object right now" workflow is still Luma Genie. But for the middle 80% of interactive 3D work — indie games, jams, AR, product viz, education — Meshy is the default answer in 2026.
The honest recommendation: sign up for the Free tier, produce five refined assets, and commit to Pro if the outputs land close enough to what you need that a Blender pass finishes them. That is the workflow the tool is built for, and it is a real workflow that ships products.