mirror of
https://github.com/langbot-app/LangBot.git
synced 2026-10-30 04:15:45 +08:00
70 lines
3.7 KiB
Plaintext
70 lines
3.7 KiB
Plaintext
---
|
|
title: "Models"
|
|
description: "Configure LLM providers in LangBot: OpenAI GPT, Anthropic Claude, Google Gemini, DeepSeek, Qwen, Ollama and OpenAI-compatible models."
|
|
---
|
|
|
|
## LangBot Models
|
|
|
|
LangBot Models is an official model service provided by LangBot. When you initialize a local instance with your LangBot Space account, available models will be automatically added to your instance without any configuration. You will receive a certain amount of free quota to get started quickly.
|
|
|
|
<img src="/images/zh/deploy/models/langbot_models.png" alt="langbot_models" />
|
|
|
|
For specific available models, please check [LangBot Space](https://space.langbot.app/models).
|
|
|
|
## Custom Models
|
|
|
|
You can also add models from other sources.
|
|
|
|
### LLM Model
|
|
|
|
Models will be used by pipelines to process messages. The first model you configure will be set as the model for the default pipeline.
|
|
|
|
|
|
|
|
You can add multiple models, and then select which model to use in the pipeline.
|
|
|
|
<img width="400px" src="/images/zh/deploy/models/model_config.png" alt="model_config" />
|
|
|
|
Enter these four parameters: `Model Name`, `Model Provider`, `Request URL`, and `API Key`, then submit.
|
|
|
|
For model capabilities, enable the ones the underlying model actually supports:
|
|
|
|
- **Vision**: enable to recognize images
|
|
- **Tool Use** (function calling / tool use): enable so the LLM can call tools during a conversation
|
|
|
|
<Note>
|
|
**Tool Use must be enabled on the model whenever you want to use tools from MCP servers, sandbox built-ins (`exec`/`read`/`write`/`edit`/`glob`/`grep`), or Skills.** Without it, even if the pipeline binds those extensions, the Agent won't call any tools.
|
|
|
|
The underlying model has to actually support function calling for this checkbox to take effect — check the model provider's docs if unsure.
|
|
</Note>
|
|
|
|
### Embedding Model
|
|
|
|
Embedding models are used to compute vector representations of messages. If you need to use knowledge bases, please configure this model.
|
|
|
|
<img width="400px" src="/images/zh/deploy/models/embedding_model.png" alt="embedding_model" />
|
|
|
|
Enter these four parameters: `Model Name`, `Model Provider`, `Request URL`, and `API Key`, then submit. After that, please configure the knowledge base to use this model.
|
|
|
|
### Using the seekdb Built-in Embedding Model (Optional Dependency Required)
|
|
The model needs no request parameters. Use `uv sync --extra seekdb` for source deployments or `uvx --from 'langbot[seekdb]@latest' langbot` for package deployments. The official Docker/Kubernetes image already includes the SeekDB dependency. Native deployment support depends on the available `pylibseekdb` wheel.
|
|
- On the "Embedding Model" page, select "seekdb-built-in";
|
|
- Click "Save" to use it immediately;
|
|
- Then select this model in your knowledge base to take effect.
|
|
|
|
### Using Chroma Built-in Embedding Model (Zero Configuration)
|
|
The system has integrated Chroma's built-in embedding model (all-MiniLM-L6-v2), no parameters required.
|
|
- On the "Embedding Model" page, select "chroma-built-in";
|
|
- Click "Save" to use it immediately;
|
|
- Then select this model in your knowledge base to take effect.
|
|
|
|
### Rerank Model
|
|
|
|
Rerank models re-score RAG retrieval results to significantly improve the precision of knowledge base answers. When a knowledge base returns multiple candidate documents, the rerank model uses a cross-encoder to precisely score each document's relevance to the query, selecting only the most relevant content.
|
|
|
|
Enter `Model Name` and select `Model Provider`, then submit. After that, select this model in your pipeline's "AI" configuration.
|
|
|
|
<Tip>
|
|
Rerank models are optional. Without one, knowledge base results are used in their original retrieval order. With a rerank model configured, results are re-scored and only the top-K most relevant results are kept.
|
|
</Tip>
|