ProjectCanItBeMade?
RevA — Pre-release
DateJune 2026
ScaleNTS
Inspection in progress

Can it be made?

The manufacturability benchmark for generative AI. Text, image, and 3D models are graded against real design-for-manufacturing rules — injection molding, CNC machining, and FDM printing — with scores checked by geometry, not vibes.

benchmark drawing no. 001 — rev a
46
Test cases
3
Modalities
3
Processes
9
Models queued

Leaderboard

QC inspection record
ModelTextImage3DOverallStatus
Claude Sonnet 4.6(Claude Code CLI, June 2026)———··———Awaiting first article
GPT-5.5(Codex CLI, xhigh reasoning, June 2026)———··———Awaiting first article
Shap-E(Hugging Face demo, June 2026)··56.96/10 tasks graded56.9In inspection
ChatGPT(free web, June 2026)———··———Awaiting first article
Claude(free web, June 2026)———··———Awaiting first article
Gemini Flash(free web, June 2026)———··———Awaiting first article
Ideogram(free tier, June 2026)·———·———Awaiting first article
Leonardo(free tier, June 2026)·———·———Awaiting first article
Copilot image gen(free, June 2026)·———·———Awaiting first article
Meshy(free tier, June 2026)··——————Awaiting first article
Tripo(free tier, June 2026)··0.02/10 tasks graded0.0In inspection
TRELLIS(public demo, June 2026)··——————Awaiting first article

Scores are 0–100 per modality; overall is the mean of graded modalities. Collection is ongoing — partial coverage is labeled per cell, and text/image rows appear once human rubric grading completes. Every leaderboard entry names the exact tier and month tested.