My third message to Meshy 3D Agent was "Rig it and make it idle and fly." Meshy did not start a job. It replied that its automatic rigging tool "is designed specifically for bipedal humanoid … (two arms, two legs, upright posture)", that my cute baby dragon "is a quadruped creature with wings and a tail", and that I could download the model and rig it in Blender. Meshy charged 0 credits for that answer and gave me 0 animation clips. Snapshapy's 3D Agent got the same sentence, rigged its own dragon in 42 s, and attached three clips to one game-sized file without another click.
I tested Meshy 3D Agent at meshy.ai on a paid Meshy Pro account and Snapshapy's 3D Agent at snapshapy.com on a paid account. Both got the same three messages, typed verbatim, and whenever a choice card appeared I picked the first option. Every number here comes from a balance I watched, a job card, a stopwatch, or a file I downloaded and opened myself.
I ran Meshy on 2026-11-03. I ran Snapshapy on 2026-11-04, a day later, after Snapshapy shipped an update.
Messages 1 and 2 asked for four concepts of a cute baby dragon for an indie game and a textured 3D model of option 1. This article is about message 3, the one that decides whether the dragon ever leaves the ground.
My verdict in 60 seconds — Meshy 3D Agent 56/100, Snapshapy 91/100
Meshy 3D Agent built me a clean dragon and then told me, politely and for free, that making it move was a job for Blender, while Snapshapy rigged it and handed me idle and flight motion in one file.
- Motion from "idle and fly" — Meshy 1/10 · Snapshapy 9/10 — 0 clips vs 3 clips (Idle A 3.12 s, Flying Forward 0.5 s, Glide 0.79 s)
- Handling a non-humanoid creature — Meshy 3/10 · Snapshapy 9/10 — a pointer to Blender vs a finished rig in 42 s
- Warning before charging — Meshy 8/10 · Snapshapy 9/10 — a clear 0-credit refusal vs a 3-credit refund promise stated before the job ran
- Engine-ready animated file — Meshy 1/10 · Snapshapy 9/10 — none vs 20,840 triangles, 5.56 MB, 1024 px textures
- Credits for the whole run — Meshy 4/10 · Snapshapy 9/10 — 39 vs 14 credits, 2.8×, $0.78 vs $0.67
- Concepts delivered — Meshy 3/10 · Snapshapy 8/10 — 1 drawn image vs 4, asked for 4
- Extra clicks beyond 3 messages — Meshy 4/10 · Snapshapy 10/10 — 4 vs 0
- Price visible before a job runs — Meshy 8/10 · Snapshapy 8/10 — Meshy's Est. Cost 9 and 30 cards vs Snapshapy's "Up to 10" spend limit
The concrete problems on the Meshy side:
- Asked to rig and animate a winged quadruped, Meshy's agent declined because its auto-rig targets bipedal humanoids, and left me with 0 clips.
- The 39 credits Meshy did charge bought a static dragon, not a moving one.
- "4 concept options" came back as four text lines on a card; Meshy drew only the one I picked, for 9 credits.
- Reaching even the static Meshy model took 4 extra clicks: a choice, a Submit and two Approves.
For an animated indie game creature, I would use Snapshapy.
A quadruped with wings and a tail: Meshy's 0-credit answer
Meshy's reply to message 3 is the most honest screen in the Meshy session. No job card appeared, and Meshy's balance read 751 before and after. The agent named the shape its tool expects, named the shape my dragon has, and offered a route: rig it in Blender.
Meshy's reply to "rig it and make it idle and fly": auto-rig is for bipedal humanoids, try Blender.
Meshy's refusal cost 0 credits and explained the reason clearly. It also ended this request inside Meshy 3D Agent: I asked for an idle and a flight and got a static file. I tested how this Meshy agent answered one winged quadruped; I am not claiming Meshy cannot animate anything.
- Meshy, message 3 — 0 credits ($0) · 0 clips · 0 animated files · 1 pointer to Blender
- Snapshapy, message 3 — 3 credits (about $0.14 on Starter) · 3 clips · 1 animated GLB
Meshy 3D Agent's chat home before the test.
Snapshapy said the rig could fail, then rigged it
Snapshapy's 3D Agent answered the same sentence with a plan and a caveat in one reply: "Rigging is running (1–3 minutes), set up light for the engine at ~20,000 triangles with 1024 px textures — 3 credits." It said Idle A, Flying Forward and Glide would be applied automatically, all in one animated GLB, "so no further step from you."
Then the line that matters for a dragon: "If the rigging engine doesn't read the dragon as two-legged humanoid, it will fail and refund the 3 credits". Same risk Meshy named, but Snapshapy tried the job with a refund on the table.
Snapshapy's reply to "rig it": rig running at ~20,000 triangles, Idle A / Flying Forward / Glide will apply automatically, refund if it fails.
It did not fail. The rig took 42 s, including Snapshapy's own reduce to about 21k triangles, and the animation 5 s more.
- Risk named before charging — Meshy yes, then declined · Snapshapy yes, then ran with a 3-credit refund promise
- Outcome — Meshy 0 clips · Snapshapy 3 clips
- Rig + animation time — Meshy not run · Snapshapy 42 s + 5 s = 47 s
Both agents saw the same problem. Only Snapshapy turned it into a dragon that idles and flies.
Idle A, Flying Forward, Glide: 4.41 seconds of motion in one GLB
About 4 s after the rig card, an animation card appeared on its own: "Idle A, Flying Forward, Glide … 3 clips". I sent nothing and clicked nothing. The rig card reads "reduced to 21k for rigging · animated automatically", and the chat header reads "14 spent".
Snapshapy rig card ("reduced to 21k for rigging · animated automatically") and the 3-clip animation card; header "14 spent".
All three clips live in one file: one import, three states for a game engine.
- Idle A — 3.12 s
- Flying Forward — 0.5 s
- Glide — 0.79 s
- Total motion — 3.12 + 0.5 + 0.79 = 4.41 s, in 1 GLB, for 3 credits
- Meshy equivalent — 0 s of motion, 0 GLBs with clips
The two flight clips are very short loops, and when I posed them frame by frame they hold one forward-leaning flying pose with no visible wing flap.
Frames I posed offline from Snapshapy's animated GLB: Idle A, Flying Forward and Glide.
Even so, 4.41 s of motion beats the 0 s Meshy's agent offered for this creature.
20,840 triangles at 1024 px, offered before I asked
Snapshapy raised the game budget one message early, in its reply to message 2: "since it's for an indie game I'd suggest a lighter rig around 20,000 triangles with 1024 px textures so it drops straight into the engine." None of my messages mentioned a triangle count.
Snapshapy's reply to message 2: 3 credits, 5 to 8 minutes, suggests a ~20,000-triangle rig with 1024 px textures.
I opened the animated GLB myself: 20,840 triangles, 5.56 MB, three 1024 px textures. That is the light version it promised, within 840 triangles of the number it quoted.
My offline render of Snapshapy's animated GLB (20,840 triangles, 5.56 MB).
- Animated file — Meshy none · Snapshapy 20,840 triangles, 5.56 MB
- Against Meshy's static GLB — 50.9 MB vs 5.56 MB, about 9.2× the bytes for a dragon that does not move
- Triangles — Meshy static 1,076,848 · Snapshapy animated 20,840 · about 52× fewer
- Texture size — Meshy's approval card set 4096 · Snapshapy animated 1024 px, 16× fewer pixels per texture
Wireframe of Snapshapy's animated GLB: a light game mesh.
In this run, the only route Meshy's agent offered to a moving dragon was a Blender session I would do by hand, at a cost no Meshy card shows.
Four clicks to reach a dragon that stands still
Meshy 3D Agent works through cards. Message 1 produced a choice card; I clicked option 1 and then Submit. Then a Text to Image approval card, Est. Cost 9, and I clicked Approve. Message 2 produced an Image to 3D approval card (Meshy 7.1, Standard, texture size 4096, Est. Cost 30), and I clicked Approve again.
Meshy image-to-3D approval card: Meshy 7.1, Standard, texture 4096, Est. Cost 30.
- Extra clicks — Meshy 4 (choice select + Submit, 2 × Approve) · Snapshapy 0
- Approval cards — Meshy 2 · Snapshapy 0
Showing the price first is a fair Meshy habit, and the scorecard credits it. The catch: all four Meshy clicks led to a static model, and none to an idle or a flight. Snapshapy used 0 extra clicks and stayed under its "Up to 10" credit spend limit.
The Sprout Dragon Meshy did build
Credit where it is due: Meshy's 3D step produced one clean dragon, 1,076,848 triangles in a 50.9 MB GLB with 3 textures.
My offline render of Meshy's dragon (1,076,848 triangles, 50.9 MB).
Meshy's static model is lighter than Snapshapy's (50.9 MB vs 79.7 MB) and very close to its concept, with leaf wings, flower-bud horns and a clover tail. Meshy's image-to-3D finished in ≈107 s against Snapshapy's 288 s, and its total job time was ≈132 s against ≈382 s, partly because Meshy never ran step 3.
Snapshapy static dragon, front view: matches concept 1.
Snapshapy's static GLB measured 1,499,980 triangles, 79.7 MB and 3 textures at 4096 px. For this job it was a starting point: the animated file has about 72× fewer triangles.
- Static GLB — Meshy 1,076,848 triangles, 50.9 MB · Snapshapy 1,499,980 triangles, 79.7 MB
- What came after it — Meshy a Blender suggestion · Snapshapy a rigged 5.56 MB GLB with 3 clips
Side notes on messages 1 and 2: one drawing, 39 credits
Concepts: I asked for 4. Meshy's agent listed four directions as text plus Custom and drew only the one I picked. At 9 credits per image, all four Meshy concepts would have been 36 credits, more than its 30-credit 3D step.
Meshy's reply to "4 concept options": four dragon directions as text, plus Custom; nothing drawn.
Snapshapy drew 4 images for 8 credits (2 each) in 38–47 s, each a single dragon in a clearly different design, though number 2 reads slender and adult and number 4 fierce and armored rather than cute.
Snapshapy's four concepts in a 2×2 grid at full resolution.
Credits: Meshy went 790 → 781 → 751 → 751, so 9 + 30 + 0 = 39 credits, or $0.78 on Pro at $20 per 1,000. Snapshapy charged 8 + 3 + 3 = 14 credits, matching its header "14 spent", or $0.67 on Starter at $19 per 400.
Meshy's 3D model ready after ≈107 s; balance 751.
- Whole run — Meshy 39 credits · Snapshapy 14 · 2.8×
Five flying creatures a month
Here is the same run priced for five flying creatures a month.
Snapshapy's 3D card Ready (288 s); header "11 spent".
- Credits — Meshy 195 · Snapshapy 70
- Dollars at the credit rate — Meshy $3.90 · Snapshapy $3.35 on Starter
- Extra clicks — Meshy 20 · Snapshapy 0
- Animated creatures — Meshy 0 · Snapshapy 5
- Clips — Meshy 0 · Snapshapy 15, 22.05 s of motion in total
- Per month on the plan's credits — Meshy 25 dragons on Pro (1,000) · Snapshapy 28 on Starter (400), 42 on Pro (600)
The dollar gap is small, $0.55 across five. The output gap is not: Meshy's $3.90 buys five static creatures and five trips to Blender, Snapshapy's $3.35 buys five that already idle and fly.
Final verdict: Meshy 3D Agent scores 50/100 for an animated indie game creature
The job was a baby dragon that can idle and fly in a game. Meshy 3D Agent delivered one drawn concept, one static 1,076,848-triangle model, 39 credits on the bill, 4 extra clicks, and 0 seconds of motion. Its answer to the rig request was clear and free, and it was still a no: this agent's auto-rig is built for bipedal humanoids, and my dragon was sent to Blender.
Snapshapy charged 14 credits, warned me up front that the rig could fail and refund 3 credits, then rigged the dragon in 42 s and attached Idle A, Flying Forward and Glide to one 20,840-triangle, 5.56 MB GLB. One tool finished the request. The other stopped, honestly and at no charge, at the step that mattered most, and a game creature that cannot idle or fly scores 50 here, not 56.
Which one I would use for this job
For a non-humanoid game creature that has to idle and fly, I would use Snapshapy. It made the one deliverable this job asked for, a rigged dragon with idle and flight clips in a single game-sized file, for 14 credits and 0 extra clicks. Plans and the 3D Agent are at snapshapy.com.
If you are happy to rig in Blender yourself, Meshy Pro at $20 still gives you a clean static model with its price shown on every card.