This release adds 1 notable feature for engineering teams evaluating rollout.
Published 2d
Productivity & Wikis
✓ No known CVEs patched
✓ No known CVEs patched in this version
Summary
AI summaryStalled on-device writing workers are killed and replaced, returning raw transcript at cleanup deadline.
Full changelog
What changed
- Isolates the on-device writing model in a persistent child process so a native MLX stall cannot hold the final result indefinitely.
- Returns the raw transcript at the existing cleanup deadline, kills only the stalled writing worker, and warms a replacement automatically.
- Preserves the existing Whisper/Qwen models, prompt cache, formatting behavior, streaming path, modes, language handling, and raw-history fallback.
- Adds separate model-cleanup, cleanup-wall, and stop-to-final measurements to local Voice Intelligence.
Quality benchmark: 23/23 exact-model cases passed with every prompt-cache hit.
SHA-256: ce7237d45332755e1d07ead8cae08482f5598cad5bd823c96a63407d46e7eeb9
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About Velora
All releases →Beta — feedback welcome: [email protected]