Skip to content

v0.6.0

Choose a tag to compare

@github-actions github-actions released this 10 Jun 19:02
· 2073 commits to master since this release
62121b8

Provider v0.6.0 (Swift CLI)

Binary hash: 8147b8ada7beca98533ce109b833458c9df6dd03334c49dc0540ae98eac6de2b
Bundle hash: 3d223fa53a3bf46b67722cc5cc0bf84637998a63bb6c1be17d19fee6ed430613
Metallib hash: 198488eb61359e953580a9c4530400feee1a06dd2f28a930a6ffa58aec66a597 (mlx==0.31.1)
Signed by: Developer ID Application: Eigen Labs, Inc. (SLDQ2GJ6TL)
Notarized: yes
Min macOS: 14.0

Install

curl -fsSL https://api.darkbloom.dev/install.sh | bash

What's Changed

  • Rebuild provider dashboard as a fleet-ops console by @Gajesh2007 in #279
  • feat(enroll): CMS-sign MDM enrollment profile + rebrand to Darkbloom by @Gajesh2007 in #281
  • feat(provider): show Terms of Service notice on darkbloom start by @Gajesh2007 in #284
  • feat(attestation): APNs-based provider code-identity attestation (v0.6.0) by @Gajesh2007 in #286
  • feat(kv-cache): encrypted SSD KV cache — primitives, verification, and flag-gated engine integration by @anupsv in #237
  • fix(coordinator): close provider WebSocket on registry removal (kill zombie connections) by @Gajesh2007 in #289
  • feat(provider): serve image + video (VLM) inference via Gemma 4 by @Gajesh2007 in #293
  • feat(coordinator): report video input modality from a video capability by @Gajesh2007 in #296
  • feat(console-ui): image upload in chat for vision models (Gemma 4) by @Gajesh2007 in #295
  • fix(coordinator+provider): kill false "model swap" untrusts on busy multi-model providers by @Gajesh2007 in #297
  • feat(coordinator): admin-gated /data state-export for EigenCloud→GCP migration (DAR-70) by @Gajesh2007 in #294
  • Background model prefetch + zero-downtime hot-swap (DAR-225) by @Gajesh2007 in #283
  • fix(coordinator): never list a build behind a public alias (hide retired builds) by @Gajesh2007 in #299
  • feat(provider): darkbloom restart + graceful download-first auto-update by @Gajesh2007 in #292
  • fix: guard graceful update drain and binary-hash rollout by @Gajesh2007 in #298
  • feat(kv-cache): sliding 5-minute SSD TTL + per-server --local cache scope by @anupsv in #290
  • v0.6.0 launch: APNs grace, vision routing, 4-bit cutover, graceful update by @Gajesh2007 in #303
  • fix: report reasoning_tokens separately in chat-completions streaming usage by @Gajesh2007 in #304
  • fix: handle two streaming-usage edge cases from Codex review (follow-up to #304) by @Gajesh2007 in #305

Full Changelog: v0.5.16...v0.6.0