
by Cloudflare
Cloudflare AI Gateway sits between your application and AI model providers. It adds analytics, logging, caching, rate limiting, routing with fallbacks and safety controls, and takes one line of code to start.
APIUsage-based
Step 1: Create a gateway in the AI Gateway section of the Cloudflare dashboard.
Step 2: Point your application's AI provider requests at the gateway, which takes one line of code to change.
Step 3: Authenticate the gateway and either send provider keys with requests, store them in the gateway, or use Unified Billing credits.
Step 4: Turn on features such as caching, rate limiting, spend limits or routing fallbacks in the gateway settings.
Step 5: Review analytics and logs in the dashboard, and use custom metadata to filter by user or team.
No, caching is disabled by default. You can enable it for the whole gateway with a default TTL, or opt in and override settings per request with headers. A cf-aig-cache-status header shows whether a response was a HIT or MISS.
By default the cache key is built from the provider, endpoint, model, authentication header and the full request body, so only exact matches hit the cache. A custom cache key header overrides this.
A provider key sent on the request is used first, then a key stored in the gateway, then Unified Billing credits if fallback is allowed. Turning on Require provider credentials prevents the Unified Billing fallback.
It lets clients pay for inference on select open models from a stablecoin wallet using the x402 protocol, with no prepaid balance. You must be based in the United States and have a credit card on file.
NEAR AI CloudNEAR AI Cloud is a private AI inference platform that runs open-weight models inside hardware enclaves and returns a signed proof with each response. It uses an OpenAI-compatible API.
Aikido SecurityAikido Security is a platform that scans code, cloud, containers and running apps for vulnerabilities, then opens pull requests with fixes. It also offers autonomous pentesting and device and runtime protection.
Phala CloudPhala Cloud runs Docker workloads, agents and private LLM inference inside hardware-backed trusted execution environments. It produces attestation reports so you can verify what ran.