The short answer
The official developer console for Moonshot AI’s Kimi API is the Kimi API Platform console. Use it to create and manage API keys. For production model selection, check the platform’s current model list and match the listed model’s capabilities and context length to your workload; the API documentation recommends checking availability before creating a chat completion.
For a new workload involving long-context work, visual understanding, coding, or knowledge work, kimi-k3.1-pro is the documented flagship starting point. If faster output is the priority, the official model guide describes kimi-k3.1-pro-highspeed as having the same capabilities at about 2–3× the speed. Confirm that either exact ID is available to your API account before deploying it.
What the console is for
The Kimi API Platform console is the developer access point for API keys—not the model-inference endpoint itself. Moonshot’s API overview says keys are created and managed in the console; requests are sent to the API using the appropriate base_url and API format. Keep the key secret: the documentation warns against exposing it in client-side code, public repositories, or logs. (API overview)
How to choose a model for production
| Workload priority | Model to evaluate | Documented reason | Check before deployment |
|---|---|---|---|
| Long-context work, visual understanding, coding, or knowledge work | kimi-k3.1-pro |
The model guide identifies it as the K3.1 flagship, with native visual understanding and a 1M-token context window; it cites coding, 3D creation, and financial research and modeling among areas of improvement. | Confirm the model appears in your account’s live model list and check the exact request limits and required capabilities for your API route. |
| Faster output while retaining the K3.1 Pro capability profile | kimi-k3.1-pro-highspeed |
The official guide calls it the high-speed variant, with the same capabilities at about 2–3× the speed. | Verify availability for your account, then assess latency and usage cost on representative requests. |
| An existing workload already using the previous-generation K3 flagship | kimi-k3 |
The model guide describes it as the previous-generation flagship, with native visual understanding and a 1M-token context window, designed for long-horizon coding, knowledge work, and deep reasoning. | Compare it with the K3.1 models on your workload; do not assume the previous-generation model is available to every account or route. |
These are documented workload matches, not a guarantee that a model will perform best for your application. The model guide describes K3.1 Pro as the current flagship and the high-speed variant’s relative speed; it does not provide a workload-specific benchmark for your prompts. For text chat, tool use, streaming, or image input, the Chat Completions documentation describes those API features, but check the live model list for model-specific capability flags. The model-list endpoint returns model IDs, context lengths, and capability flags, and is the relevant availability check.
Minimum path to an API request
- Open the Kimi API Platform console and create an API key. Store it as a server-side secret, such as an environment variable; do not expose it in browser-side code, public repositories, or logs. (API overview)
- Choose the API format and configure its
base_url. The official quickstart describes choosing a model and making requests over HTTP or with an OpenAI or Anthropic SDK. - Before creating a completion, query the model list and select an available model ID whose capability flags and context length fit the request.
The API supports multiple request formats, so follow the documentation for the specific route you choose rather than assuming parameters are interchangeable.
Sources
Kimi API Platform: API Overview
Model List - Kimi API Platform
List Models - Kimi API Platform