OpenClaw 2026.6.10-beta.2: fast-talks auto mode beta, GLM routing fixes
Small beta — 11 merged PRs — previewing automatic fast mode for short conversational turns, plus Z.AI/GLM routing corrections and a fix keeping trusted tool policies intact when hook registries are composed. Superseded by stable 2026.6.10.
PUBLISHED2026-06-22
OBSERVED2026-08-11
AGE1mo
SOURCES1
- Fast mode can engage automatically for short conversational turns and return to normal mode for longer runs, with bounded fallback and delivery behaviour (#85104)
- Zhipu GLM overload responses classified as overloaded so failover picks the right path (#93241)
- Synthesized GLM-5 models fall back to the manifest base URL instead of misrouting (#94461)
/thinkmenu levels resolved from the runtime catalog for live-discovered models (#94067); GLM-5.2 reasoning levels exposed (#94136)- Composed hook registries retain trusted tool policies needed by approval-sensitive flows (#94545)
- Stale per-channel origin fields reset on channel switch (#95328); cron delivery awareness stays attached to the target session (#93580)
COMMUNITY
No curated reactions recorded for this event. Facts and takes are kept in separate layers — community context is added by hand, never blended into the record above.