Short answer: Meshy’s 3D Agent markets a chat path from idea/photo to print-ready export (STL / 3MF into a slicer), including a public “97% slicer pass rate” claim from Meshy’s internal test. Useful when you want conversation → cleanup → slicer handoff in one suite. If your job is one photo → one inspectable GLB draft before you trust a print, start with inspection, not a print-ready promise. On trellis-2.com the Mini pack is $4.99 / 150 credits = one 512³ job (1024³ / 1536³ = 300 / 500 credits). Credits are one-time and never expire; failed jobs return reserved credits; GLB + PBR included. And be clear: trellis-2.com is an independent hosted TRELLIS.2 service — not Microsoft, not the official project site. Structure contrast only; don’t copy Agent sentences. Show the math; don’t dunk.
Cost-per-accepted-GLB deep-dive: Hosted TRELLIS.2 vs Meshy: Cost Per Accepted GLB.
Two honest use cases
| You are… | Better fit |
|---|---|
| Want chat → concepts → print-ready STL/3MF in one Agent flow | Meshy 3D Agent print/slicer path |
| Need animation, multi-format suite, team seats | Meshy (or similar) monthly suite |
| One photo, one draft GLB, inspect before print | Checklist + Mini $4.99 |
| First-ever hosted TRELLIS.2 taste test | Mini, then decide |
| You already know the subject is clean and only need a reviewable mesh | Skip chat; run 512³ |
No dunking: print-ready Agent marketing is a real product shape for people who want conversation + slicer export. An inspect-first hosted generator is a real product shape for priced drafts you refuse until the mesh passes QA. Different jobs. “Print-ready” is a claim to verify in your slicer — not a substitute for manifold / wall / scale checks.
First: what trellis-2.com is (and is not)
- Independent third-party image-to-3D host with its own billing and support.
- Not Microsoft, not affiliated with Microsoft, not the official TRELLIS.2 project site.
- Microsoft publishes open-source TRELLIS.2 weights; this site sells managed generation, GLB + PBR download, and credit packs.
If you need official Microsoft product support, this is the wrong counter. If you need a browser path from one photo to a downloadable GLB with a printed credit cost — then a slicer test after inspection — this is the product under review.
Live numbers (2026-10-27 Asia/Shanghai)
trellis-2.com/pricing
| Output | Credits reserved |
|---|---|
| 512³ | 150 |
| 1024³ | 300 |
| 1536³ | 500 |
| Pack | Price | Credits | Max 512³ |
|---|---|---|---|
| Mini | $4.99 | 150 | 1 |
| Starter | $12.90 | 300 | 2 (or 1×1024³) |
| Pro | $38 | 3,000 | 20 |
| Studio | $129 | larger pool | on-page |
One-time packs never expire. Failed jobs return reserved credits. GLB + PBR included on published packs.
Mini math: $4.99 = one accepted 512³ slot if the job succeeds and you keep the mesh after inspection. A successful-but-ugly result still spends credits — that is why the inspect-first checklist exists before you open a slicer.
Meshy 3D Agent (public feature page — structure only)
Public page shape (paraphrase, not copied sentences):
- Describe idea / photo / sketch in chat.
- Approve 2D concepts, then convert to 3D.
- Export for pipeline — including print paths (STL / 3MF / OBJ into named slicers) and a marketed slicer pass-rate figure from Meshy’s internal test.
Treat that pass-rate as Meshy’s internal marketing metric, not your printer’s guarantee. Always run your slicer preview and a small test print. Agent chat may be free on Meshy’s Free plan; generation actions still spend credits — verify in-app. Monthly pools ≠ Mini’s single forever slot.
Inspect-first checklist (before you trust a print)
A) Input (before you spend 150 credits)
- Single clear subject; readable silhouette.
- Decent lighting on surfaces that will print.
- No tiny-text / logo reliance — fine print rarely survives as geometry.
- Expected back side — camera never saw it ⇒ don’t expect a perfect reverse.
- Start at 512³ for the first draft; save 1024³ / 1536³ for after accept rate is known.
B) Mesh acceptance (before slicer)
- Silhouette — orbit; fused limbs, melted handles, floaters, holes in the back.
- Thin walls — paper-thin parts that will break or Z-fight.
- Manifold / normals — open in Blender or a repair tool; fix non-manifold edges before STL.
- Scale — measure a known real-world edge.
- UV / PBR sanity — only if textures matter for the print (painted resin, etc.).
- Export opens — GLB loads in site viewer and in your DCC.
C) Print gate (after GLB accept)
- Convert / export STL or 3MF only after B passes.
- Slice in your slicer (Bambu / Orca / Creality / etc.) — watch overhangs, supports, wall thickness.
- One small resin or FDM test beats three blind regenerations.
- Only then spend another 150 / 300 / 500 on a retry or resolution step-up.
Failures on trellis-2.com: Activity shows the reason; reserved credits return; retry is a new run. Do not treat “job succeeded” as “print-ready.”
Print-ready Agent vs Mini $4.99 — which CTA fits?
| If you… | Path |
|---|---|
| Want chat + marketed print-ready / slicer handoff | Evaluate Meshy Agent on their credit rules and run your slicer anyway |
| Want one photo → inspectable GLB → then decide print | Mini $4.99 |
| Want one retry at 512³ or one step to 1024³ | Starter $12.90 |
| Batch ~20 accepted 512³ SKUs | Pro $38 |
CTA
For inspect-first, one 512³ draft: Mini $4.99 / 150 credits. Orbit the GLB. Run the checklist. Slice only after accept. For chat-to-print Agent workflows, compare Meshy’s Agent page on meshy.ai/features/3d-agent with the same adult energy — no trash talk, and no substitute for your own slicer test.
Also see: Hosted TRELLIS.2 vs Meshy: Cost Per Accepted GLB.
Sources (live, 2026-10-27 Asia/Shanghai)
- https://trellis-2.com/pricing — Mini $4.99/150cr = 1×512³; 150/300/500 for 512/1024/1536; forever one-time; failed return credits
- https://www.meshy.ai/features/3d-agent — Agent structure; print-ready / slicer export narrative; internal slicer pass-rate marketing
- https://trellis-2.com/blog/trellis-2-vs-meshy-accepted-glb — sibling cost post
UNVERIFIED / notes
- Exact Meshy credits per Agent print-path action: in-app — not invented.
- Meshy 97% slicer pass rate: Meshy internal test marketing — not your printer’s guarantee; do not treat as independent benchmark.
- Meshy pricing may localize currency — no FX invention.
- Search volume / CTR: unavailable — not invented.
- Independent host disclaimer: trellis-2.com is not Microsoft.
