Feature Request
Moonshot AI released Kimi K3 as open source on July 27, 2026. Since aiproxy already supports multi-tenant AI gateway routing with intelligent channel selection and load balancing, K3 would be a valuable addition to the channel lineup.
Why K3 makes sense for aiproxy
aiproxy is the Sealos ecosystem's production-grade AI gateway with features like smart retry, priority channel selection, and MCP support. Adding K3 as a channel fits naturally:
- OpenAI-compatible API: K3 follows the standard
/v1/chat/completions protocol, which means it can be added as another OpenAI-compatible channel in aiproxy's plugin-based architecture -- essentially a config entry with the K3 base URL and model name
- Chinese model coverage: K3 strengthens aiproxy's coverage of Chinese LLMs, which is important for Sealos ecosystem users who need both domestic and international model access
- Multi-channel load balancing: aiproxy's smart retry + priority channel selection means K3 can be mixed into routing pools alongside other models, providing more options for cost optimization and failover
- Self-hosted scenario: Since K3 is open source, Sealos users could deploy their own K3 instances and register them as private channels through aiproxy, keeping data within their infrastructure
Technical implementation
Adding K3 should be straightforward -- it would follow the existing OpenAI-compatible channel registration pattern. The configuration would specify the K3 base URL, API key handling, and model name mappings.
Context
We recently completed a K3 review and setup via API gateway through TeamoRouter's enterprise platform. The model's OpenAI-compatible interface makes channel integration clean. For aiproxy users in the Sealos ecosystem, having K3 as a supported channel would expand their routing options with a capable open source model.
Feature Request
Moonshot AI released Kimi K3 as open source on July 27, 2026. Since aiproxy already supports multi-tenant AI gateway routing with intelligent channel selection and load balancing, K3 would be a valuable addition to the channel lineup.
Why K3 makes sense for aiproxy
aiproxy is the Sealos ecosystem's production-grade AI gateway with features like smart retry, priority channel selection, and MCP support. Adding K3 as a channel fits naturally:
/v1/chat/completionsprotocol, which means it can be added as another OpenAI-compatible channel in aiproxy's plugin-based architecture -- essentially a config entry with the K3 base URL and model nameTechnical implementation
Adding K3 should be straightforward -- it would follow the existing OpenAI-compatible channel registration pattern. The configuration would specify the K3 base URL, API key handling, and model name mappings.
Context
We recently completed a K3 review and setup via API gateway through TeamoRouter's enterprise platform. The model's OpenAI-compatible interface makes channel integration clean. For aiproxy users in the Sealos ecosystem, having K3 as a supported channel would expand their routing options with a capable open source model.