Changelog
Continuous deployment updates, model catalog additions, and platform enhancements.
Architectural Enhancements
Asynchronous Rust data plane, isolated mTLS Cortex sidecar for zero-retention guardrails, and tenant-pinned Postgres transactions.
API & Wire Updates
Full parity with OpenAI `/v1/chat/completions`, `/v1/embeddings`, Anthropic Messages, and live `/capabilities` runtime introspection.
Gateway Performance
Zero-copy streaming SSE pipelines, connection pooling across 8+ providers, and 15,000+ sustained requests/sec per container node.
Model Catalog Additions
Instant access to frontier LLMs (OpenAI, Anthropic Claude 3.5, Google Gemini 2.0, DeepSeek) at exact provider list prices with zero markup.
- v1.0.1Release
v1.0.1 — Capability discovery + honest 501s
Endpoints that aren't yet wired now return a clear HTTP 501 with a roadmap pointer instead of a confusing upstream error. New /capabilities endpoint lets SDKs introspect what's live.
#capabilities#api#reliabilityRead release notes → - v1.0Release
v1.0 — Public Launch
nRouter v1.0 is live — managed LLM gateway with the full Google Vertex AI lineup, full dashboard, credit-based billing, AI guardrails, prompt management, A/B testing, team-scoped budgets, and observability — every feature on every tier.
#launch#v1.0Read release notes → - v0.8Release
v0.8 — Observability: logs, callbacks, alerts, data policy
Per-org logging callbacks (Langfuse, Datadog, S3, Slack), budget and spend alerts, full request-log search, and four data-policy modes so you decide what request content we retain.
#observability#logging#alertsRead release notes → - v0.7Release
v0.7 — Model catalog with leaderboard rankings
Browse every registered model on one page. Sort by latency, cost, or quality. Switch models with a one-line env var change — your code stays exactly the same.
#models#catalog#leaderboardRead release notes → - v0.6Release
v0.6 — GDPR data subject requests
Self-service GDPR access, erasure, withdraw-consent, and consent records. Every administrative action lands in the audit trail. SOC 2 controls are active. HIPAA-eligible with a BAA on Enterprise.
#gdpr#compliance#privacy#soc2Read release notes → - v0.5Release
v0.5 — Credit safety: reserve + settle
Every LLM request now reserves credits before the call, settles against the real cost after, and releases the hold on every failure path. Negative balances are blocked at the database — not by application logic.
#credits#billing#safetyRead release notes → - v0.4Release
v0.4 — Team management
Invite teammates, assign roles (owner / org_admin / member / viewer), and see per-key spend by member. Single-org-per-user enforced at the database; one team per user within that org, by design.
#team#rbac#invitationsRead release notes → - v0.3Release
v0.3 — Usage analytics dashboard
Per-model, per-key, per-team spend charts on a single page. Filter by date range, export CSV, drill into outliers. Numbers render in Geist Mono — easy on the eyes during the monthly close.
#analytics#usage#reportingRead release notes → - v0.2Release
v0.2 — Prompt management with A/B testing
Server-side prompt templates with versioning and Jinja2 variables, plus deterministic traffic-split A/B tests. Same hash, same variant — every single time.
#prompts#ab-testing#templatesRead release notes → - v0.1Release
v0.1 — Guardrails: pre/post-call enforcement
Guardrails in the request path on every plan — PII redaction, prompt-injection detection, keyword & regex filtering, custom webhook checks, and post-call response scanning — scoped at org, team, and key level.
#guardrails#security#piiRead release notes →
Release Lifecycle & Governance
Release Cadence & Versioning Policy
Our release engineering practices are designed for mission-critical production workloads. Every gateway deployment adheres to strict semantic versioning standards, executes as a zero-downtime rolling update, and provides forward-compatibility guarantees.
Semantic Versioning (SemVer 2.0.0)
Releases strictly adhere to semantic versioning standards (MAJOR.MINOR.PATCH). Breaking schema or contract changes occur exclusively in major version increments with a mandatory 6-month deprecation horizon. Minor releases add backward-compatible capabilities, and patch releases ship bug fixes and internal performance tuning.
Zero-Downtime Rolling Deploys
Every release is packaged into immutable container digests and promoted through staging test gates before reaching production. Rolling updates preserve in-flight streaming connections and gracefully drain active requests before decommissioning older container instances—guaranteeing zero dropped tokens.
Forward-Compatibility & Wire Guarantees
nRouter guarantees strict backward and forward wire compatibility with the OpenAI and Anthropic API specifications. Client SDKs, LangChain, LlamaIndex, and custom HTTP integrations continue operating reliably across releases without requiring code modifications or downtime.
Dynamic Catalog & Zero-Restart Sync
Upstream foundation model catalog additions, pricing updates, and context ceiling adjustments synchronize dynamically within 30 to 60 seconds from our distributed configuration store. New models become available instantly across our global fleet without requiring gateway restarts.
Release Inquiries
Frequently asked questions about nRouter release notes
How frequently does nRouter ship gateway releases?
We continuously deliver patch updates, performance optimizations, and upstream provider model integrations multiple times per week. Minor and major platform capabilities are documented in structured release notes.
Do platform releases introduce gateway downtime?
No. All production deployments execute as zero-downtime rolling updates. In-flight streaming requests complete uninterrupted, and new requests are seamlessly routed to healthy container instances.
How do client SDKs maintain compatibility across releases?
Our gateway maintains strict backward compatibility with the OpenAI API wire specification. Client libraries and custom HTTP integrations continue working without modification across minor and patch releases.
Where can developers verify what endpoints and capabilities are live?
Developers can introspect live capabilities programmatically via our authenticated /capabilities endpoint or review current model offerings and latency benchmarks in the public catalog at /models.