← News
Release

qm v0.1.14

QM 0.1.14 refines approvals and goals, adds the newest models, and gets faster under load. Newest models: Claude Sonnet 5.5 with adaptive thinking, GPT-6.1 Sol, an Astra Ultrafast speed switch, and…

Originally published on github.com →

QM 0.1.14 refines approvals and goals, adds the newest models, and gets faster under load.

  • Newest models: Claude Sonnet 5.5 with adaptive thinking, GPT-6.1 Sol, an Astra Ultrafast speed switch, and automatic fallback when a gateway model refuses or is unavailable.
  • Approvals got stricter and clearer: the approval card is now the only way to approve credential use, strict-posture approvals show the gated call's arguments, and single-use credential grants are atomic again.
  • Security screening simplifies to three modes: off, observe, and enforce. Memory adds provenance-aware disclosure and audience isolation.
  • App sharing is one General access menu (Restricted, your org, or anyone with the link), and sharing outside the org stays off unless an org admin enables it.
  • Goals are simpler and harder to game: exact timed floors, file-based evidence, a visible indicator, and only a human stop or an independent verifier can end one.
  • Subagents feel present: their activity shows in the parent session, sessions keep working while they run, elapsed time ticks live, and opt-in Slack task cards track ongoing work.
  • Long conversations got lighter: transcripts page by turns, model-only context stays out of view, and older history loads during live turns.
  • Faster under load: sandbox operations no longer queue behind one machine lock, startup skips session-counter locks, and prompt inventories stay stable for cache reuse.
  • Dependency security: fastify and hono updated to clear five new advisories, including an auth bypass.

What's Changed

Full Changelog: https://github.com/yc-software/qm/compare/v0.1.13...v0.1.14