feat(ai-security-guard): replace denyMessage with structured DenyResponseBody (#3642)

fix: remove non-standard thinking/reasoning_max_tokens from Claude-to… (#3667)

Vertex structured outputs (#3649)

feat(provider): 优化 Azure multipart 处理 || feat(provider): Optimize Azure multipart processing (#3651)

fix: Remove unused pilot values in the helm chart (#3657)

fix: Fix a bug in template processor occurred when mixing default and non-default namespaces in one config (#3652)

support gemini & claude domain setting (#3638)

docs: replace broken higress.io links (#3644)

fix(kingress): include header match in duplicate-route dedup key (#3580)

fix: update kingress status logic (#3575)

ai-statistics: increase default value_length_limit and emit AILog during streaming usage (#3624)

feat: add CI workflow to build and push plugin-server image on release (#3632)

Update model configurations for new GPT-5.4 versions

feat: Add responses to the default path suffix list of model-router and model-mapper (#3585)

chore(nacos): upgrade nacos-sdk-go/v2 from v2.3.2 to v2.3.5 (#3587)