One gateway for
every AI call
in your org.
Ingate sits between your applications and LLM providers. Get complete observability, prompt versioning, automated evals, and provider fallback, with a single URL change.
// Before: direct provider call
const client = new OpenAI({ apiKey: "sk-..." });
// After: one line swap
const client = new OpenAI({
baseURL: "https://api.ingateai.com",
defaultHeaders: { "X-Ingate-Provider": "openai" }
});
// That's it. Zero code changes to your app logic.
// Ingate handles logging, evals, routing, fallback.Trusted by engineering teams at
Everything your AI stack needs.
A single layer between your applications and LLM providers that gives you visibility, control, and resilience across every AI interaction.
Transparent Proxy
Route any LLM request through Ingate with a single header. Supports OpenAI, Anthropic, Azure, Ollama, and any HTTP-based provider. Sub-millisecond internal overhead with full SSE streaming.
Full Observability
Every request logged asynchronously with smart extraction of model, tokens, latency, and cost. Per-user session tracking, Prometheus metrics, usage analytics, and a built-in dashboard.
Prompt Registry & Playground
Version-controlled prompt templates with namespace organization. Interactive playground for testing prompts with real-time streaming and side-by-side model comparison.
Eval Engine & Datasets
14 built-in evaluators plus versioned datasets for systematic testing. Import test cases from CSV/JSONL, sample from production logs, and run batch evaluations against any dataset version.
Multi-Tenancy & RBAC
Full multi-org platform with teams, apps, and scoped API keys. Four-level role hierarchy (viewer → owner) with per-resource access control. Key rotation with zero-downtime overlap periods.
Cost Tracking & Budgets
Automatic dollar cost calculation from token counts with per-model pricing. Set budgets at org, app, or key level with soft alerts and hard caps. Usage breakdown by provider, model, team, or user.
Security & Guardrails
PII redaction and prompt injection detection on live requests. Audit log of every admin action. Webhooks with HMAC signing. Response caching for deterministic requests. Provider keys encrypted with AES-256.
Flexible Ingestion
Beyond the proxy: REST ingestion API for async/sidecar architectures, OpenTelemetry OTLP/HTTP receiver for instrumented apps, and BYOS (Bring Your Own Storage) for complete data sovereignty.
Simple, honest pricing.
Start free with logging and basic proxy. Unlock the full platform when you're ready for production.
Free
For developers evaluating Ingate or running side projects.
- Transparent proxy to all providers
- Full SSE streaming support
- Request & response logging
- 7-day log retention
- Built-in dashboard
- Community support
- Up to 5 API keys per org
- Prompt registry & versioning
- Eval engine (14 evaluators)
- Provider fallback chains
- Per-key rate limiting & scopes
- PII redaction & guardrails
- Cost tracking & budget controls
- Audit log & webhooks
- Response caching
- Datasets & playground
- BYOS (Bring Your Own Storage)
- Custom retention policies
- Priority support & SLA
Enterprise
For teams shipping AI to production. Everything, unlocked.
- Transparent proxy to all providers
- Full SSE streaming support
- Request & response logging
- Configurable log retention (30/60/90/∞ days)
- Built-in dashboard & Prometheus metrics
- Prompt registry & versioning
- Eval engine: 14 evaluators + LLM Judge
- Provider fallback chains
- Format translation (OpenAI ↔ Anthropic)
- Up to 100 API keys, scopes & per-key rate limiting
- PII redaction, guardrails & injection detection
- Cost tracking & budget controls (org/app/key)
- Audit log & webhooks (HMAC-signed)
- Response caching (Redis-backed)
- Datasets, playground & model comparison
- BYOS (Bring Your Own Storage)
- REST ingestion API & OTel receiver
- Session & user tracking analytics
- RBAC (viewer/member/admin/owner)
- 99.9% uptime SLA
- Priority support & dedicated onboarding