Supported Providers
- OpenAI
- Google Gemini
- Anthropic Claude
- Novita AI
- OpenRouter
- Google Vertex AI
- Amazon Bedrock
- Azure OpenAI
- OpenAI-compatible
- Anthropic-compatible
Best for: Latest GPT models and reliable performance.Get an API key:
- Visit OpenAI API Keys
- Create a new key for Kodus
- Add billing information
- Include the API version, as in
https://your-gateway.example.com/v1. Kodus calls<Base URL>/responses, and lists models from<Base URL>/models. - If the gateway only serves chat completions, connect it as OpenAI-compatible instead.
- The URL must use
https://and resolve to a public address, because Kodus sends your key there. Gateways on a self-hosted install’s private network are not supported. - Requests go out with
store: false, so neither the gateway nor OpenAI is asked to keep them. - Use OpenAI’s own model ids (for example
gpt-5.4). A model the gateway renames is not recognized as a reasoning model, so its reasoning level is not sent. - Not supported here: gateways that need their own headers besides the key (Portkey, Cloudflare AI Gateway), and Azure OpenAI, which has its own Azure OpenAI provider.
Connecting Kimi (Moonshot) and GLM (Z.ai)
Moonshot and Z.ai each offer a subscription plan on a different endpoint than their pay-per-token Developer API. Each plan is a separate account with its own key — pick the base URL that matches the key you have. The Moonshot tile uses the Developer API’s Anthropic-compatible endpoint (https://api.moonshot.ai/anthropic). The Z.ai tile uses https://api.z.ai/api/anthropic, which Z.AI documents for its GLM Coding Plan. To use the Z.AI Developer API, choose OpenAI-compatible under Custom and use https://api.z.ai/api/paas/v4/. For a Kimi Code Plan key, choose a compatible provider under Custom and enter the base URL from the table.
- Kimi (Moonshot AI)
- GLM (Z.ai)
For Kimi Code Plan, start with a low cap and adjust it for your account. See Tuning concurrency. Kimi K2.7 Code always reasons. Kodus hides its temperature field and omits temperature from requests. See Temperature.
Pinning OpenRouter providers
OpenRouter is a router — when you request a model (e.g.moonshotai/kimi-k2), it forwards the call to one of several upstream providers (Moonshot direct, Together, Groq, Fireworks, Novita…). Each call can land on a different backend. That’s convenient, but it introduces silent variance:
- Quality drift — upstreams run different precisions (FP8, INT4, full) and give subtly different outputs for identical prompts
- Tool-calling inconsistency — some backends don’t support function calling the same way, leading to malformed tool use
- Reasoning format variance — one upstream honors
reasoning_effort, another onlythinking.enabled, another ignores both - Latency swings — p50 can jump from 800ms to 4s between calls as routing changes
- Rate-limit surprises — you hit quota on a backend you didn’t explicitly choose
How to pin
When your BYOK provider is OpenRouter, the Advanced settings panel shows an OpenRouter routing section with two fields:- Pin providers (in order) — comma-separated list of upstream names (e.g.
moonshot, together). OpenRouter tries them in order and uses the first available. - Allow fallbacks — when off, requests hard-fail if none of the pinned providers are available. When on (default), OpenRouter can fall back to any other upstream that serves the model.
Advanced: raw JSON override
If you need fields beyondorder and allow_fallbacks (e.g. ignore, data_collection, require_parameters), switch Thinking / Reasoning to Custom in Advanced settings (Custom JSON override) and paste the full routing payload — it’s merged into providerOptions alongside any reasoning config: