Cloudflare launches AI Gateway 2.0 with agent routing and cost controls

Read the original article →

What happened

Cloudflare launched AI Gateway 2.0 on July 26 with agent-aware routing, automatic model fallbacks, per-team cost budgets, and prompt caching across OpenAI, Anthropic, Google, and open-model providers.

The gateway adds observability, rate limiting, and DLP scanning in front of LLM API calls. Cloudflare said the gateway processed more than 10 billion tokens daily in beta. It is available on Pro, Business, and Enterprise plans.

Why it matters

Teams running multiple models and agents need a control plane for cost, routing, and data loss prevention. A gateway at the CDN edge reduces latency compared with routing through a separate proxy.

Automatic fallbacks also improve reliability when a primary model is down or rate-limited.

MintedBrain take

An AI gateway is worth adding once you run more than one model provider in production. Start with cost budgets and logging before enabling DLP rules that might block legitimate agent tool calls.

References

This article was originally published at Cloudflare docs. For the full piece, read the original article.

Discussion

  • Loading…

← Back to News