diff --git a/nodes/src/nodes/llm_kimi/doc.md b/nodes/src/nodes/llm_kimi/doc.md new file mode 100644 index 000000000..02de4f68f --- /dev/null +++ b/nodes/src/nodes/llm_kimi/doc.md @@ -0,0 +1,83 @@ +--- +title: Kimi (Moonshot) +date: 2026-06-11 +sidebar_position: 1 +--- + + + Kimi (Moonshot) - RocketRide Documentation + + +## What it does + +Connects [Moonshot AI](https://www.moonshot.ai/)'s Kimi models to your pipeline via the Moonshot cloud API. Used primarily as an `llm` invoke connection by agents and other nodes that need an LLM. Can also be used directly via lanes. + +The Moonshot API is OpenAI-compatible, so this node uses the OpenAI SDK / `langchain-openai` client pointed at the configured base URL. + +**Lanes:** + +| Lane in | Lane out | Description | +| ----------- | --------- | ---------------------------------------------------- | +| `questions` | `answers` | Send a question directly, receive a generated answer | + +## Configuration + +| Field | Description | +| --------------- | ---------------------------------------------------------------------------------------------------------------- | +| Model | Kimi/Moonshot model to use (see profiles below) | +| API Key | Moonshot API key (`sk-` prefixed) | +| Server base URL | Moonshot endpoint (default `https://api.moonshot.ai/v1`; use `https://api.moonshot.cn/v1` for the China platform) | + +## Profiles + +| Profile | Model | Context | +| --------------------- | ------------------ | ----------- | +| Kimi K2.6 _(default)_ | `kimi-k2.6` | 256K tokens | +| Kimi K2.5 | `kimi-k2.5` | 256K tokens | +| Moonshot v1 8K | `moonshot-v1-8k` | 8K tokens | +| Moonshot v1 32K | `moonshot-v1-32k` | 32K tokens | +| Moonshot v1 128K | `moonshot-v1-128k` | 128K tokens | + +The Kimi K2 family are reasoning models with a 256K context window and strong agentic performance; the classic Moonshot v1 models are cost-effective options across 8K, 32K, and 128K context sizes. + +> **Temperature is model-aware.** Kimi K2 reasoning models require `temperature=1` and reject other values with a 400 error, so the node forces it for those profiles; Moonshot v1 models default to `0`. + +> Moonshot also offers `moonshot-v1-{8k,32k,128k}-vision-preview` image-input models. Those belong in a dedicated vision node (see the `llm_vision_*` nodes), not this text LLM node, which exposes only the `questions → answers` lane. + +## Upstream docs + +- [Moonshot AI / Kimi API documentation](https://platform.moonshot.ai/docs) + +## Reference + + + + +| Property | Value | +| --- | --- | +| Class type | llm | +| Capabilities | invoke | +| Protocol | `llm_kimi://` | + +**Data lanes** + +| Input | Produces | +| --- | --- | +| `questions` | `answers` | + +**Profiles** + +| Profile | Title | Model | +| --- | --- | --- | +| `kimi-k2-6` | Kimi K2.6 | kimi-k2.6 | +| `kimi-k2-5` | Kimi K2.5 | kimi-k2.5 | +| `moonshot-v1-8k` | Moonshot v1 8K | moonshot-v1-8k | +| `moonshot-v1-32k` | Moonshot v1 32K | moonshot-v1-32k | +| `moonshot-v1-128k` | Moonshot v1 128K | moonshot-v1-128k | + +**Configuration sections** + +| Section | Fields | +| --- | --- | +| Kimi (Moonshot) | `kimi.profile` | +