Hi Vercel team,
I'm looking for help diagnosing a reproducible Preview deployment failure on our Next.js project.
The application build completes successfully, including dependency installation, compilation, TypeScript checks, static page generation, and creation of /vercel/output.
Immediately afterward, during:
"Deploying outputs..."
the deployment transitions to ERROR with:
"The deployment failed because of an internal Vercel error."
There is no specific public error code.
Project details:
Project: flowark Project ID: prj_AQsoDjzPpsUdL5rY53ZVQ9AwVqym Team: FLOWARK Team ID: team_QdtDm14NfWuqoQcs3e0ZGDG9 Branch: feature/onboarding-v1-clean Framework: Next.js 16.3.4 Node.js: 24.x
The original failing commit was:
8b0b96b54702bb0648a97c4b9e3a0bd5cdcf560d
It failed three times at the same post-build stage:
dpl_69edMrCWrGLtHkZiJktVzpVpqNow dpl_3NTnmHeZW4e8HBn1gt14B5XtkLhj dpl_2ETFyBbUxNLtDVYudt8QgsCVkGsu
The previous successful Preview deployment was:
dpl_5S46yRJFEUtKWRqnfvrKE8ZMDH6E SHA: 77eac7c4feeae313db39d90098e31fa7e6b38098
We first contacted the Vercel Support Agent. It inspected the deployment metadata and reported that:
- the Next.js build completed successfully
- the failure occurred during output packaging/upload
- the successful deployment registered 11 Serverless Function outputs
- the failed deployments registered no outputs
- no Hobby plan quota or concurrency restriction was triggered
- the new tesseract.js / tesseract.js-core WASM/binary assets were suspected of causing the output packager to fail
Based on that diagnosis, we changed the packaging configuration to isolate the OCR/Tesseract runtime assets instead of broadly tracing them across API routes.
The packaging correction was committed as:
0834b94
Local validation after that change passed:
- 3,440 tests passed
- 40 existing skips
- TypeScript passed
- production build passed
- ESLint: 0 errors
- two clean 205-migration database replays passed
- a 77-manifest trace audit passed
- OCR/Tesseract tracing was isolated from unrelated API routes
However, the first and only Preview deployment of the corrected commit still failed at exactly the same point:
Build Completed in /vercel/output → Deploying outputs... → internal Vercel error
We intentionally did not keep retrying or make additional speculative source changes.
At this point we would really appreciate help from someone at Vercel who can inspect the backend/internal publish trace.
Specifically, could you help determine whether the failure is occurring during:
- Serverless Function packaging/serialization
- artifact upload/storage
- /vercel/output validation
- WASM/native asset processing
- function registration
- or another internal publication step?
The public deployment logs do not expose the failing sub-stage, so we cannot safely determine what to change next.
Importantly, we are not asking for a redeploy or a generic troubleshooting checklist. The builds consistently succeed; we are trying to identify the exact post-build publication failure before making any further source changes.
Thank you very much for taking a look.