Same Gemini models.
Billed at or below list.
SayGM serves Google’s models through the same API shape you already call, priced at or below Google’s own list price on every token. Keep your SDK, change the base URL and key, and the same key serves Claude, GPT and Gemini.
At or below list price,
by design.
Independent providers compete to serve each request on SayGM, and the winning bid sets your rate. Every token line is capped at Google’s published list price, so any discount applies to input, output and cache reads alike.
You pay the rate of the provider that served the request, never above list. Funding your balance costs exactly the amount you fund it with.
A gateway you can verify yourself.
Calling Google through SayGM adds one hop, and that hop is sealed. The SayGM gateway runs inside an Intel TDX trusted execution environment, so its operators and the host machine cannot read your prompt, and the attestation quote proving it is independently verifiable. Optional guardrails redact personal data and secrets inside that enclave before a request leaves it, and confidential open-weight models keep a prompt sealed end to end on the same key.
On a Gemini model, the provider serving the request still receives your prompt and processes it under its API terms. If you need a data-handling contract of your own with Google, sign it directly.
Verifiable by anyone
Remote attestation lets you confirm for yourself exactly what code the gateway is running.Sealed from the routing layer too
Most privacy claims stop at the model. This one covers the gateway itself.Built for the audit trail
Pairs with signed logs of which model handled what, for teams that need to show their work.What you could save.
Pick a model, set your monthly volume, see the delta.Most input tokens in real traffic are cache reads, and both gateways price them well below a fresh one — so the share matters more to the bill than the totals do. The default is what SayGM measures across its own traffic.
Price per million tokens.
USD per 1M tokens, uncached input / output at the standard tier. SayGM’s side is the best rate a provider offered, as of 19 minutes ago. Google’s is its published list price.SayGM vs. Gemini API, line by line.
- Work that can wait: the Batch API is 50% off list, and Flex inference is 50% off for latency-tolerant work.
- Zero data retention in your own name for eligible features, set up through Google Cloud per its data governance docs.
Common questions about switching.
Is SayGM cheaper than the Gemini API?
Across the Gemini models this page compares, 15% below Google’s list price for the middle one, at the best rate a provider is bidding right now. You pay the rate of the provider that serves each request, never above list. For work that can wait, Google’s Batch API is 50% off list.
Does the Gemini SDK work unchanged?
Yes. SayGM serves generateContent and streamGenerateContent under the same model ids, so the Gemini SDK works with SayGM’s base URL and key. The same key also serves Claude and GPT.
Does long context cost the same?
Google prices some models higher above 200k prompt tokens, as its pricing page shows. The table above compares the standard rate; SayGM’s cap at list applies to each rate it bills.
Can I use Batch or Flex through SayGM?
SayGM serves real-time requests at the standard tier. For jobs that can wait, Google’s Batch API is 50% off list with a 24-hour target turnaround, and Flex inference trades latency for a lower price. You can run real-time traffic through SayGM and batch jobs direct.
What happens to my data?
Your request passes through the SayGM gateway, sealed in an Intel TDX enclave, and reaches Gemini from the serving provider’s account, under that account’s API terms. Google Cloud customers can set up zero data retention for eligible features, with steps and exceptions (such as grounding with Search) set out in its data governance docs. If your workload needs those terms in your own name, buy direct.
Can I switch back?
Yes. Point the SDK back at Google and your code runs as before. SayGM has no subscription to cancel.