Kyma API gives your app one key for 100+ models across chat, embeddings, image, and audio, with published prices and measured uptime for every model.
Keep the OpenAI SDK you already use and change the base URL and key:
from openai import OpenAI
client = OpenAI(base_url="https://kymaapi.com/v1", api_key="YOUR_KYMA_KEY")
print(client.chat.completions.create(model="qwen-3.6-plus", messages=[{"role": "user", "content": "Hello!"}]).choices[0].message.content)With the Anthropic SDK, set base_url to https://kymaapi.com and call messages.create as usual. New accounts get $0.50 of free credit with no card required, and it can be spent on the free-tier models. More setup options are in the docs quickstart.
- One API key and one balance for every model.
- A published price for each model, listed on Pricing.
- Uptime measured for each model, published on Status.
- Public model rankings built from real customer usage, on Rankings.
- If a model's route fails, requests fall back automatically.
Kyma API works with OpenClaw, Cursor, Claude Code, Cline, Roo Code, Continue, Aider, OpenCode, and other OpenAI- or Anthropic-compatible clients.
- MCP: connect any MCP client to
https://mcp.kymaapi.com/mcp. You sign in with OAuth, and each connection has a spend cap you set. - Agent Skills: run
npx skills add kyma-api/kyma-skillsto add four skills to your agent: connect a tool, pick a model by price and uptime, check spend and credits, and send a test completion.
| Repository | What it contains |
|---|---|
kyma-mcp-plugin |
The MCP server guide, the stdio bridge, and plugin manifests for coding agents. |
kyma-skills |
Agent Skills: connect-my-tool, pick-a-model, check-spend-and-credits, send-a-test-completion. |
openclaw-provider |
The plugin that adds Kyma API models to OpenClaw. |
Questions: hello@kymaapi.com
