Say gm tofrontier models.
Drop-in access to Claude, GPT, and Gemini.
Your prompts routed through a verifiable TEE.
sign up with email or Google — start instantly, no card required
Easy integrations
Point Cursor, Cline, Claude Code, or your AI SDK at the gm gateway. One base URL. No rewrites.
VS Code → Cline settings → OpenAI Compatible
API ProviderOpenAI CompatibleBase URLhttps://api.saygm.com/v1API Keygm_live_…Modelclaude-sonnet-4-6Access you can build on.
See live pricingNo surprise cutoffs
models stay reachable
No silent price hikes
rates published, not slipped in
No breaking changes
your integration keeps working
Your prompt reaches the provider — and no one else in between.
Externally hosted models run on hardware the provider controls, so the provider processes your prompt under its own API terms. What gm adds: the route is sealed inside a TEE, so the operators, the host, and gm never see your prompt on the way — only the provider does.
Not even the maker sees your prompt.
Open models gm runs inside the TEE itself, so your prompt is decrypted only within the enclave — invisible to the maker, the host, and us. The strongest tier: nothing leaves the box in the clear.
Verified by hardware, not by policy.
The gateway runs in an Intel TDX enclave. The chip itself signs a measurement of exactly what's running — so the proof comes from Intel's silicon, not from a promise on our website.
Reading hardware-signed evidence…
Popular models, live pricing.
The best available pricing across every provider serving each model, per million tokens in USD.