scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-22 00:31:44 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 51

docs(release-notes): rewrite key highlights and add guardrail narrative section

Browse source
This commit is contained in:
Ishaan Jaffer 2026-02-21 14:58:48 -08:00
parent e0ad538de1
commit 789c74e806
1 changed files with 19 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

26
docs/my-website/release_notes/v1.81.14.md
View file

@ -42,16 +42,28 @@ pip install litellm==1.81.14
## Key Highlights
- **Use Claude Sonnet 4.6 on day 0** — [reasoning, computer use, prompt caching, and 200K context, working across Anthropic and Vertex AI from the moment it launched](../../docs/providers/anthropic)
- **Guardrail Model Garden** — [Easily browse available guardrails — competitor blockers, GDPR PII, EU AI Act, prompt injection, and more — and deploy in one click](../../docs/proxy/guardrails/policy_templates)
- **3 new built-in LiteLLM guardrails** — [competitor name blocker, topic blocker, and insults filter — all configurable, no external API required](../../docs/proxy/guardrails)
- **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)
- **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
- **Guardrail Model Garden** - [Browse and deploy production-ready guardrail policies in one click — competitor blockers, topic blockers, keyword filters, GDPR PII, EU AI Act, prompt injection, and more](../../docs/proxy/guardrails/policy_templates)
- **Block topics, competitors, and keywords at the gateway — no external API** - [3 new built-in guardrails: competitor name blocker, topic blocker, and insults/keyword filter. Define your own lists, apply to specific teams or keys, swap providers without changing code](../../docs/proxy/guardrails)
- **See what your guardrails are actually doing in production** - [Guardrail tracing UI shows policy hits, detection method, and match details per request — know exactly when and why a guardrail fired](../../docs/proxy/guardrails)
- **Test guardrail policies before you ship them** - [Upload a CSV dataset to the compliance playground and validate against real traffic. Get AI-generated policy suggestions with latency estimates before anything goes live](../../docs/proxy/guardrails/policy_templates)
- **Claude Sonnet 4.6 — day 0** - [Full support across Anthropic and Vertex AI: reasoning, computer use, prompt caching, 200K context](../../docs/providers/anthropic)
- **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.
## Guardrail Model Garden
The biggest theme in this release is operationalizing guardrails at scale. The question isn't just "can I turn on a guardrail" — it's "how do I know if my guardrail is working, what happens when it's over-sensitive, and how do I fix it in production without a full redeploy."
This release ships three things to address that:
1. **Guardrail Model Garden** — a searchable, categorized library of pre-built policies you can deploy in one click. Competitor blockers, topic filters, keyword lists, GDPR/EU AI Act compliance templates, prompt injection detection, and more. Start from a template and customize, instead of building from scratch.
2. **3 new built-in guardrails** — competitor name blocker, topic blocker (keyword + embedding-based), and insults/keyword filter. All run at the gateway level, configurable per-team or per-key, no external API call required. If you're on AWS Bedrock today and want to add a layer of topic blocking without routing traffic to another service, these work out of the box.
3. **Guardrail tracing** — every request that hits a guardrail now shows the policy name, detection method, and exact match in the logs view. When a policy is firing too aggressively in production, you can see exactly which rule triggered and on what input — instead of guessing.
---
---