OpenClaw 2026.6.8: GLM-5.2 and Claude Haiku 4.5 support, key-free search now opt-in
Stable release adding GLM-5.2 and Claude Haiku 4.5 catalog entries with normalized provider-qualified model IDs, rich Telegram delivery, and WhatsApp ACP binding support. Operators relying on automatic fallback to key-free web search providers must now opt in explicitly; admin privileges are also required for HTTP session/model override surfaces.
PUBLISHED2026-06-16
OBSERVED2026-08-11
AGE1mo
SOURCES1
- Models: GLM-5.2 support and Claude Haiku 4.5 catalog entries; provider-qualified model IDs normalized across OpenRouter and Google Vertex paths; managed SecretRef auth; bounded model browsing; safer OpenAI/Anthropic tool-schema recovery
- Web search behavior change: key-free providers (Parallel Free, DuckDuckGo, Ollama, Codex Hosted Search) stay explicit opt-ins instead of being selected automatically when no API-backed provider is configured
- Channels: Telegram renders structured text with tables, lists, expandable blockquotes, preserved intentional line breaks and CLI-backed replies; WhatsApp honors configured ACP bindings; channel chunking fixed around surrogate pairs and Infinity limits
- Usage footers: built-in full footer rendering, default templates, per-turn usage state, credential-aware limits, fixed-decimal formatting, warnings on broken templates
- Agent/Gateway runtime: active main sessions marked before restart shutdown aborts, yielded subagent runs paused on terminal abort, trusted subagent thinking overrides clamped through provider/model fallback, reset archive fallback reads restored, heartbeat events de-duplicated, session identity exposed in runtime prompts
- Gemini CLI runs in an isolated runtime home on the selected OpenClaw OAuth/API-key profile, preventing ambient Google machine credentials from overriding it
- Discord: generated auto-thread titles get a 60-second timeout and 4,096-token reasoning output budget, clamped to the selected model output cap
- Memory/state: oversized OpenAI embedding batches split before 431s, QMD search available in transient mode, SQLite avoids WAL on NFS volumes, full reindexes preserve rollback/cache recovery
- Security: admin privileges required for HTTP session/model override surfaces; unknown OpenAI agent selectors rejected
- UI/mobile: workspace files start collapsed, WebChat backscroll survives streaming, desktop session picker stays interactive, iOS reconnects stale foreground Gateways
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.