2025 v2.5.1
ai-job-search-dev/ai-job-searchFeb 6, 2026 → May 26, 2026109 days between releases
What's New
New release introduces Dream Talk portal, bootstrap resume support, and lifecycle hardening.
New Features
- Bootstrap full-model downloads now preserve partial files and support resume
- Dream Talk owner-portal adds streamed SSE replies, TTS streaming, and mobile routing
- Hermes local-provider calls now use longer request timeouts for slow backends
- Lifecycle hardening includes compose health waits and delayed port reuse
- LAN web guidance now points to proxy surfaces instead of raw API ports
Improvements
- Release documentation, security policies, and contribution guidelines expanded
- Bundle service CPU limits and fallback model serving added
- Stale cloud compose-cache invalidation implemented
- CLI roadmap documentation added
Bug Fixes
- Fixed Strix Halo timeout cascade with extended request timeouts
- Bootstrap downloads now cap progress display at 100%
- Recover cleanly on next start/restart after failed downloads
Upgrade Notes
The provided release notes are for different repositories (speech-to-speech vs Dream Server), so upgrade guidance cannot be provided.
The Two Releases
2025 release
Feb 6, 2026· 9 months ago
A large rollup release for Hugging Face speech-to-speech consolidating the project's foundational development. It spans Dockerization, MPS support, multilingual pipelines, multiple ASR and TTS backends, and a REST API.
Full release notesDream Server 2.5.1 - Fresh Install & Bootstrap Recovery
May 26, 2026· 5 months ago
A fresh-install and bootstrap-recovery patch release following the 2.5.0 audit hardening cycle. It hardens model download resumption, local-provider timeouts, lifecycle paths, and includes Dream Talk owner-portal functionality.
Full release notes