Skip to content

feat(inference): add Prime mode for text and video - #220

Merged
gujieye merged 2 commits into
mainfrom
feat/fast-mode-inference
Sep 22, 2026
Merged

gujieye merged 2 commits into
mainfrom
feat/fast-mode-inference

Conversation

@gujieye

@gujieye gujieye commented Sep 22, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Add --prime and --workspace-id to bl text chat and bl video generate.
  • Route Prime requests through workspace-scoped endpoints while honoring custom Base URL overrides.
  • Require an explicit model, keep video submission and polling on the same host, and limit text Prime to Chat Completions.
  • Add unit/E2E coverage and generated text/video references.
  • Keep unrelated commands, including knowledge and security, unchanged.
  • Harden FC-backed OSS publishing with 20 retries, 600-second attempt timeouts, capped exponential backoff, and fresh OIDC tokens for each FC attempt.

Validation

  • Changed-file formatting, lint, and type checks passed.
  • Prime helper tests: 3 passed.
  • Text Prime/help E2E tests: 7 passed.
  • Video Prime/help E2E tests: 5 passed.
  • FC upload OIDC refresh regression test: 1 passed.
  • git diff --check origin/main...HEAD passed.

Full-repository vp check remains blocked by 15 pre-existing formatting issues from main; none are in the Prime change set.

@gujieye
gujieye merged commit a5a68b8 into main Sep 22, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant