SI-APIPortuguês (BR)Open console

Updated · SI-API

SIAPI.AI

The only API you need.

Powerful open models. Exceptional value.

Access DeepSeek, GLM, Kimi, MiniMax and Qwen with one API key. Build applications, automate workflows or connect your coding Agent. Pay for the tokens you use, or choose a shared Coding Plan.

Explore modelsView API pricing
Chat Completions
curl "https://siapi.ai/v1/chat/completions" \
  -H "Authorization: Bearer $SIAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-flash",
    "messages": [{
      "role": "user",
      "content": "Let’s build something."
    }],
    "max_tokens": 256
  }'
Set your endpoint and API key
YOUR MODEL TOOLKITDeepSeekGLMKimiGLMMiniMax

Built for the way you build

SIAPI.AI is the home of SI-API, an independent multi-model API service. Connect applications, automation workflows and coding Agents through one account and API key, with usage and charges in one console.

Applications
Add model responses to your product, build assistants, or process text through an HTTP API or SDK-compatible client.
Workflows
Connect API requests to your existing automation. Select a model for each step and track the resulting token usage in one account.
Coding Agents
Keep your preferred coding tool and configure SI-API as its model provider. Follow the guide for your tool, operating system and interface.

SI-API provides the gateway, billing and integration guides independently of the model and tool providers.

Powerful open models

Choose from seven catalog models across five providers:

Compare the Flash models for everyday text and coding work, and evaluate the other models on your more demanding tasks. Each model page lists its API ID, token prices and available reasoning settings so you can make your own trade-off.

Text, streaming and client-defined function tools are covered by the setup guides. Image input and reasoning controls depend on the model and client; MiniMax M3.1 Flash uses model-default reasoning.

Compare model capabilities →

Exceptional value, with costs you can check

Pay-as-you-go prices separate uncached input, cached input and output. Estimate the cost of your actual workload before choosing a model, then review reported tokens and charges in Usage.

DeepSeek V4.1 Flash

1 million uncached input tokens + 1 million output tokens

$0.685

With the same total input and output, replacing half the input with reported cache hits gives an estimated cost of $0.6225.

Illustrative usage, calculated from the current shared price catalog. Excludes tax and credit rewards; actual cache hits depend on reported usage.

A lower token rate is one part of the decision. Compare model results and response times on your own task, especially when switching providers or settings.

Estimate your API cost → · View all API token prices →

One key, multiple models

Keep your SI-API base URL and key, then select a different catalog model ID in your request or Agent. Each Agent configuration includes all seven models. The selected startup model is a default; it does not remove the others. Key model restrictions and available plan or credit still apply.

For OpenAI-compatible clients, use https://siapi.ai/v1. Choose Chat Completions or Responses to match your client. For Anthropic SDKs, use https://siapi.ai as the base URL for Messages requests.

Changing a model ID does not make every model accept identical parameters. Check reasoning levels, image input and output limits when switching.

Choose your API protocol → · Moving from another provider? →

Start with your application or Agent

Building with the API

Get a key, set the base URL and send a small first request. The interactive quickstart includes cURL, Python and Node examples for Chat Completions, Responses and Messages, with a streaming option.

Open the API quickstart →

Using a coding Agent

Choose one of seven setup guides. Each includes the full model catalog, installation and launch steps, downloadable configurations, and client-specific model and reasoning-effort controls where available.

If you are migrating, match the protocol your client uses before replacing its endpoint, key and model ID. Check a small request, streaming and tool behavior before moving a larger workflow.

Follow the migration checklist →

Choose how you pay

Pay as you go

Top up API credit in USD and pay for actual token usage at the selected model’s rates. This can suit intermittent requests, application prototypes and workloads that change from week to week. A Coding Plan is not required.

Choose an API credit amount →

Optional shared Coding Plan

For recurring coding work, compare Pro, Max and Pro Max. All seven models share the allowance, with different points rates per model. Every request must fit both the 5-hour and 7-day windows. Choose a monthly subscription or a one-time purchase.

The displayed effective plan prices assume full weekly allowance use over 30 days. Lower utilization increases your effective cost; check both quota windows against your workload.

Your default auto key uses an active plan first. Enable pay-as-you-go in Billing to use credit when no plan can cover the request.

Compare Coding Plans → · Billing basics

Contact SI-API

Email contact@siapi.ai for setup or billing questions. Include the model, tool, request or order ID and time when relevant. Keep API keys and full prompts out of support messages. The troubleshooting guide covers common errors.