docs(release-notes): polish v1.81.14 - narrative paragraph, consolidated guardrail templates, merged competitor bullet

docs/my-website/release_notes/v1.81.14.md

@@ -47,9 +47,11 @@ pip install litellm==1.81.14
 - **Test guardrail policies before shipping** — [upload a CSV dataset to the compliance playground and validate policies against real traffic; get AI-generated policy suggestions with latency overhead estimates](../../docs/proxy/guardrails/policy_templates)
 - **Turn any OpenAPI spec into an MCP server** — [paste a spec and get a working MCP server instantly, via API or UI](../../docs/mcp)
 - **Call any prompt management system from a single API** — [the new Prompt Management API works with Langfuse, LangSmith, and others without requiring per-integration code](../../docs/proxy/prompt_management)
-- **Major performance batch** — 20+ targeted optimizations shipping together: callback sort moved to registration time, Pydantic serialization eliminated from the hot path, OpenAI client params pre-computed at startup, cost calculator O(n²) loops removed, and more
+- **Major performance batch** — 20+ targeted optimizations across router algorithms, logging overhead, cost calculator, and connection management — meaningfully lower latency and CPU overhead on every request
 
+---
 
+This release includes the largest single batch of performance work since v1.74. The most impactful change moves async/sync callback sorting from per-request to registration time (~30% speedup for callback-heavy deployments). On top of that: Pydantic round-trips eliminated from the logging hot path, OpenAI client init params pre-computed once at startup, quadratic deployment scan removed from usage-based routing, and several O(n²) → O(1) fixes in the router's team filter and model list lookups. Combined, these changes add up for high-throughput deployments that were hitting CPU ceilings.
 
 ---
 
@@ -300,13 +302,13 @@ pip install litellm==1.81.14
     - Guardrail tracing UI: show policy, detection method, and match details - [PR #21349](https://github.com/BerriAI/litellm/pull/21349)
 
 - **AI Policy Templates**
-    - GDPR Art. 32 EU PII Protection policy template - [PR #21340](https://github.com/BerriAI/litellm/pull/21340)
-    - EU AI Act Article 5 policy template (split into 5 dedicated sub-guardrails) - [PR #21342](https://github.com/BerriAI/litellm/pull/21342), [PR #21453](https://github.com/BerriAI/litellm/pull/21453)
-    - French language support for EU AI Act Article 5 guardrail - [PR #21427](https://github.com/BerriAI/litellm/pull/21427)
-    - Prompt injection detection policy template - [PR #21520](https://github.com/BerriAI/litellm/pull/21520)
-    - Aviation and UAE policy templates with tag-based filtering - [PR #21518](https://github.com/BerriAI/litellm/pull/21518)
-    - Airline off-topic restriction policy template - [PR #21607](https://github.com/BerriAI/litellm/pull/21607)
-    - SQL injection policy template - [PR #21806](https://github.com/BerriAI/litellm/pull/21806)
+    - Seven new ready-to-deploy policy templates ship in this release:
+        - GDPR Art. 32 EU PII Protection - [PR #21340](https://github.com/BerriAI/litellm/pull/21340)
+        - EU AI Act Article 5 (5 sub-guardrails, with French language support) - [PR #21342](https://github.com/BerriAI/litellm/pull/21342), [PR #21453](https://github.com/BerriAI/litellm/pull/21453), [PR #21427](https://github.com/BerriAI/litellm/pull/21427)
+        - Prompt injection detection - [PR #21520](https://github.com/BerriAI/litellm/pull/21520)
+        - Aviation and UAE topic filters with tag-based routing - [PR #21518](https://github.com/BerriAI/litellm/pull/21518)
+        - Airline off-topic restriction - [PR #21607](https://github.com/BerriAI/litellm/pull/21607)
+        - SQL injection - [PR #21806](https://github.com/BerriAI/litellm/pull/21806)
     - AI-powered policy template suggestions with latency overhead estimates - [PR #21589](https://github.com/BerriAI/litellm/pull/21589), [PR #21608](https://github.com/BerriAI/litellm/pull/21608), [PR #21620](https://github.com/BerriAI/litellm/pull/21620)
 
 - **Compliance Checker**
@@ -314,9 +316,8 @@ pip install litellm==1.81.14
     - CSV dataset upload to compliance playground for batch testing - [PR #21526](https://github.com/BerriAI/litellm/pull/21526)
 
 - **Built-in Guardrails**
-    - Competitor name blocker guardrail - [PR #21719](https://github.com/BerriAI/litellm/pull/21719)
-    - Competitor guardrails: streaming discovery, name variations, pre/post call split - [PR #21533](https://github.com/BerriAI/litellm/pull/21533)
-    - Topic blocker guardrail with keyword and embedding-based implementations - [PR #21713](https://github.com/BerriAI/litellm/pull/21713)
+    - Competitor name blocker: blocks by name, handles streaming, supports name variations, and splits pre/post call - [PR #21719](https://github.com/BerriAI/litellm/pull/21719), [PR #21533](https://github.com/BerriAI/litellm/pull/21533)
+    - Topic blocker with both keyword and embedding-based implementations - [PR #21713](https://github.com/BerriAI/litellm/pull/21713)
     - Insults content filter - [PR #21729](https://github.com/BerriAI/litellm/pull/21729)
     - MCP Security guardrail to block unregistered MCP servers - [PR #21429](https://github.com/BerriAI/litellm/pull/21429)