Skip to main content

​
Quick Setup

1

Get your API key

Sign up at kymaapi.com/signup. You’ll get your API key instantly; the $0.50 signup credit on the free tier is added once you verify your email (Google sign-in counts as verified).
2

Configure OpenClaw

Edit ~/.openclaw/openclaw.json:
Replace kyma-YOUR_API_KEY_HERE with your actual key.
kimi-k2.6 is outside the free tier and returns 402 model_requires_paid until your first top-up. The default, gemma-4-31b, and the other models in this block run on the signup credit.
3

Start coding

That’s it. OpenClaw will use Gemma 4 31B by default. Switch models anytime by changing the primary field.
Use model aliases like code or best. We repoint an alias by hand when a better model ships; pin a full model id if you need a model that never changes.

​
Choosing a model

  • Agentic coding: Use agent (Kimi K2.6) for autonomous tool use, after your first top-up
  • Code generation: Use code (Qwen 3 Coder) for code-focused tasks
  • Quality: Use best (Qwen 3.6 Plus) for complex reasoning
  • Budget / long context: Use cheap (DeepSeek V4 Flash)

​
Switch Models

Change the default model in your config:
Or add multiple models and switch between them in OpenClaw.

​
Auto-Failover

Kyma automatically handles provider failures:
  • If the primary infrastructure for your model is down, Kyma retries on backup infrastructure
  • Most failures are invisible — you get a response without noticing
  • Check X-Kyma-Fallback response header to know if a fallback was used

​
Troubleshooting

Make sure you’re using https://kymaapi.com/v1 (not localhost). Kyma is a cloud API — no local setup needed.
Check the model ID is correct. Use GET https://kymaapi.com/v1/models to see all available models.
Your free 0.50 credit may have been used up, or the model you called is outside the [free tier](/pricing#free-tier). Add credits at [kymaapi.com/billing](https://kymaapi.com/billing). Packages start at 5.
Try a faster model like qwen-3-32b. Larger models (DeepSeek V3, Qwen 3.6 Plus) take longer but produce higher quality output.

​
Audio Models (optional)

OpenClaw runs as your primary agent, but if you build voice-enabled tooling on the side (call summaries, transcript pipelines, voice notes), Kyma’s audio aliases drop in like any chat model:
  • transcribe → whisper-v3-turbo — 228× realtime speech-to-text
Use them via POST /v1/audio/transcriptions. Text-to-speech is documented in the Audio API reference and Model Aliases.

​
Prompt Caching

Some models cache a repeated prompt prefix. pricing.cached on /v1/models is the rate Kyma publishes for cached input on that model — usually below its input rate, but not always, so compare the two for the model you use. Every chat completion reports usage.cost, what the request was charged, and usage.cache_discount, Kyma’s figure for what caching saved on it. Whether a prefix is cached automatically or needs a cache_control marker depends on the model’s creator. See Prompt Caching for details.