kimi-k3 API: billing and integration
Compare models, explore supported endpoints and check current pricing before integrating.
Model identifier
kimi-k3 Model overview
Kimi K3 combines native visual understanding with extended software engineering and knowledge work.
Tasks to evaluate
- Evaluate a repository investigation that must retain earlier constraints.
- Cross-check a document explanation against its accompanying images.
Integration notes
The official kimi-k3 API keeps thinking enabled and requires complete assistant messages in follow-ups, including reasoning_content and tool_calls. Verify route handling.
Based on provider documentation, not a GetAPI execution test. Supported request fields and limits depend on the selected endpoint.
Official references: Moonshot AI — Kimi K3 model card · Reviewed:
Source description
No model description was supplied. Use the declared endpoint and live model settings instead of assuming capabilities.
Provider: Moonshot
Declared endpoints
Endpoint definitions vary by model; a listed route does not imply support for every parameter or tool.
- openai ·
POST /v1/chat/completions
Billing and groups
Dynamic or tiered billing
Check the current model, group and billing unit before sending a request.
Enabled groups in snapshot: Kimi
Loading...
Before your first request
- Confirm the exact model identifier and supported endpoint.
- Create a server-side API key with the required permissions and quota.
- Match the model endpoint type to a compatible request protocol before sending traffic.
A model listed by the gateway is not automatically compatible with Codex. Use its declared endpoint as a starting point, then confirm request fields and billing before testing.