Open-source platform for 3D AI agents. Turn text or a photo into a rigged, animated GLB avatar, give it an LLM brain, memory and a wallet, and embed it anywhere with one web component. Remote MCP server, x402 payments, Three.js, browser-native.
JavaScript 3d-generationai-agentsanimationavatardigital-humangenerative-aiglbgltfimage-to-3dllmmcpmcp-servermodel-context-protocolsolanatext-to-3dthreejsweb-componentswebglwebxrx402
docs(openai): note the dead TRELLIS revision and the failing vision QA gate in the resubmission checklist
nirholascf0cdcc6f196
diff --git a/prompts/store-submissions/_generated/openai-submission.md b/prompts/store-submissions/_generated/openai-submission.mdindex c5128a0b5..0b6043aa9 100644--- a/prompts/store-submissions/_generated/openai-submission.md+++ b/prompts/store-submissions/_generated/openai-submission.md@@ -55,6 +55,23 @@ current video predates the fix, and the reviewer validates test cases against itregenerated `chatgpt-app-submission.json` (`node scripts/build-openai-portal-fields.mjs --json`)and bump the version to 1.0.1.+**Two production faults found while verifying, both of which a reviewer would hit:**++- **The TRELLIS lane was dead from 2026-09-18 to 2026-09-29.** Revision `model-trellis-00028-r7b`+ was built from `c87bd4119`, which predates the libusb fix in `35d798585`, so open3d failed to+ import on every boot while the service kept accepting jobs that then sat `queued` forever. A+ refinement queued there for over twelve minutes during this verification. Traffic was moved back+ to `model-trellis-00027-tr5` (same config, loaded daily from 09-09 to 09-17) at 20:05Z on+ 2026-09-29, and it reported `ready: true` at 20:11Z. The fixed image still needs a build and+ deploy from `workers/model-trellis/cloudbuild.yaml`; after that, send traffic to latest.+- **Vision QA is failing open again (B3).** A refinement came back as a small telescope on a+ large flat slab. The geometry scorer flagged it (`planar: true`, flatness 0.16), but every+ vision rung failed: Vertex is billing-denied ("Lightning dunning decision is deny" on project+ 93741856042), the OpenAI account answers `billing_not_active`, the free OpenRouter model was+ rate-limited, and NVIDIA answered in prose rather than JSON. The slab shipped. Clearing either+ billing hold restores the gate; until then, run the test cases a few times before recording+ and keep a clean take.+### The first two rejections (2026-09-12)**Version 1.0.0 was submitted on 2026-09-12 at 00:10Z and rejected twice that day.** The portal