We earn a reward if you sign up to SayGM through our links. Your price is the same.
LessPerToken

SayGM vs OpenRouter › Kimi K3

Kimi K3 API: 39% cheaper on SayGM than OpenRouter

Kimi K3 costs $1.93 input and $9.66 output per million tokens on SayGM, against $3.17 and $15.83 on OpenRouter once its card fee is added.

Try Kimi K3 on SayGM No subscription, no top-up fee.
Where you buyInputCached inputOutput
Reference list price$3.00Not published$15.00
OpenRouter, incl. 5.5% fee†$3.17Not published$15.83
SayGM$1.93Not published$9.66

US dollars per million tokens, standard tier. “Not published” means the gateway lists no separate cache price. †OpenRouter's price depends on which provider you route to.

What would Kimi K3 cost you?

Example workloads

You save per month

$0

OpenRouter, incl. 5.5% fee
$0
SayGM
$0
Per year
$0
Get a SayGM key

Coding agents re-send long context, so most of their input is cache reads. That is why the cache slider matters more than the totals.

Is Kimi K3 private on SayGM?

The standard Kimi K3 route is sealed through SayGM's gateway, then processed by the serving provider under its API terms. SayGM also offers a confidential -tee version that keeps the prompt sealed from request to response, at a smaller discount.

Claude, GPT, Gemini and most open models

Your app→SayGM, sealed→Model provider

The model provider handles your request under its own API terms.

Confidential models (-tee)

Your app→SayGM, sealed→Model, sealed

The model runs in an enclave too, so the prompt stays sealed from request to response. Kimi, GLM, Qwen and DeepSeek have -tee versions, with smaller discounts.

Switch in two minutes

Change the base URL and key in your existing OpenAI, Anthropic or Gemini SDK code. Copy the exact Kimi K3 model ID from the models list.

Python
from openai import OpenAI

client = OpenAI(
    base_url="https://api.saygm.com/v1",
    api_key="YOUR_SAYGM_KEY",
)
resp = client.chat.completions.create(
    model="MODEL_ID",
    messages=[{"role": "user", "content": "Hello"}],
)

FAQ

What is the cheapest way to use Kimi K3 through an API?

Of the options we compare, SayGM: $1.93 input and $9.66 output per million tokens. That is 39% less than OpenRouter after its card fee.

How much would I save on Kimi K3 at 10M input and 2M output tokens a month?

About $24.68 a month: $63.30 on OpenRouter against $38.62 on SayGM.

Is Kimi K3 on SayGM private?

The standard route is sealed through SayGM's gateway, then the serving provider reads the prompt. A confidential -tee version of Kimi K3 keeps prompts sealed end to end, at a smaller discount.

Do you earn money from this page?

Yes. SayGM rewards us when someone signs up through our links and adds credit. Your price is the same either way.

Use Kimi K3 for 39% less

One URL change. Keep OpenRouter as your backup.

Try SayGM