v2026.8.1 is out. The first full release: security-reviewed, hardened, and out of beta. What's new →
Why it pays

The cheapest incident is the one that stops itself

3 arguments for putting a control point in front of your AI traffic: what it stops you spending, what it saves you assembling, and what it stops you explaining afterwards.

  • Hard budget caps
  • One tool, not five
  • OWASP LLM Top 10
Cost

What a runaway agent costs before anyone notices

An agent stuck in a retry loop bills at machine speed. Without a hard cap, the first signal is usually the invoice. Set your own assumptions below and compare the two outcomes.

Normal daily burn$267
Runaway, uncapped$19,200
Runaway, hard cap at the limit$1,600
$17,600 not spent, because the proxy returned HTTP 402 instead of forwarding

Your numbers, your assumptions. This is arithmetic on the values above, not a benchmark or a claim about your workload. What the gateway actually does is enforce the cap: monthly USD limits per team or per API key, a configurable warning threshold, and a choice between warning and hard-blocking when the limit is reached.

Consolidation

1 control point instead of five integrations

Each of these is a separate product or a separate build in most stacks. Here they share 1 policy engine, 1 audit trail, 1 identity model, and 1 dashboard.

Capability coverage. Which specific products you would otherwise buy, and what they cost, depends entirely on your stack.
What you needUsuallyHere
Prompt-injection and jailbreak filteringA guardrails vendor, per-request pricingBuilt in, plus fan-out to five providers if you want them
Token accounting and budget enforcementAn LLM gateway or a spreadsheet after the factPer team and per key, enforced at the proxy
PII and secret redactionA DLP product or a home-grown middlewareMasked outbound, restored inbound
LLM tracing and metricsA separate observability integrationLangfuse, Prometheus, OpenTelemetry, and Jaeger
MCP and agent authorizationCustom code per clientOAuth 2.1 with DCR, plus cryptographic agent identity
Inventory of what AI you actually runA spreadsheet, usually staleDiscovered in-band from live traffic
Audit evidence for a frameworkLog shipping plus a query you write yourselfStructured audit events, browsable and exportable
Risk

Mapped to the framework you report against

The OWASP LLM Top 10 for 2025, and the control in the gateway that addresses each item. Guardrails evaluation scores your deployment against this list directly.

OWASP itemControl in the gateway
LLM01 Prompt injectionPolicy engine patterns, guardrail provider fan-out, and system-prompt injection of behavioural rules
LLM02 Sensitive information disclosureOutbound PII and secret redaction with round-trip unmasking, plus canary token leak detection
LLM03 Supply chainMCP tool and A2A AgentCard rug-pull detection, the Skill Security Hub, and content-addressable approval
LLM04 Data and model poisoningShadow-AI inventory with quarantine and deny enforcement at the proxy
LLM05 Improper output handlingResponse scanning and dedicated evaluation cases for markdown exfiltration, XSS, and SQL strings
LLM06 Excessive agencyFail-closed MCP tool permissions, 3-tier hierarchy, per-tool rate limits, and agent autonomy floors
LLM07 System prompt leakageSystem prompt governance with audit logging, and extraction test cases in the evaluation corpus
LLM08 Vector and embedding weaknessesEvaluation cases for indirect injection via retrieved documents and cross-context leakage
LLM09 MisinformationContent moderation policy packs and a dedicated misinformation evaluation category
LLM10 Unbounded consumptionBudget limits with HTTP 402 enforcement, per-user and per-tool sliding-window rate limits, and circuit breakers

Put the cap in place before you need it

Budgets, redaction, and tool permissions are all opt-in. Turn on the ones you need and leave the rest alone.