Skip to main content

Model Providers

Introduction​

Model providers connect AI capabilities to your Casibase instance. They power everything from chatbots to document analysis by integrating large language models (LLMs) directly into your knowledge base workflows.

Add a Model Provider​

Adding a model provider takes just a few clicks:

Click the Providers button on the home page.

Casibase-home-providers

Click the Add button to add a model provider.

Casibase-providers-add

Fill in Model Provider Details​

Fill in the model provider details and click the Save & Exit button.

Casibase-model-openai

Supported Providers​

ProviderModels & Features
OpenAIGPT-3.5-turbo, GPT-4, o1 series. Reasoning models show step-by-step thinking. Web search available for real-time information.
Azure OpenAIAccess OpenAI models through Azure infrastructure with official SDK support.
Alibaba CloudQwen models and DeepSeek (v3, v3.1, v3.2, R1). Web search with inline results.
DeepSeekDeepSeek V3.2 with chat and reasoner variants for advanced reasoning. Updated pricing for improved affordability.
ClaudeAnthropic's Claude family including the new Opus 4.5 for complex tasks, plus Sonnet and Haiku variants for balanced performance and speed.
Moonshot (Kimi)Moonshot v1 series (8k, 32k, 128k context) and Kimi K2 models with preview, turbo, and thinking variants. Auto-tier pricing with kimi-latest.
Hugging FaceOpen-source models including Llama-2, ChatGLM2, Baichuan2, and thousands more.
OpenRouterUnified API access to Claude, PaLM, GPT-4, and other providers.
OllamaRun models like Llama, Mistral, and Phi locally on your machine. No API key needed.
Configuration Notes

When adding a provider, configure these fields:

  • Category: Choose Model for language models or Embedding for text vectorization
  • Type: Select the specific provider (OpenAI, Hugging Face, Ollama, etc.)
  • API key: Your API key or access token, required for cloud providers like OpenAI, Azure OpenAI, DeepSeek, not needed for Ollama (runs locally without authentication)

Example: Adding OpenAI

Casibase-model-select

Streaming Support

Not all models support streaming output (showing responses as they're generated). GPT-3.5-turbo and most modern models do, but if you're using older or custom models, check with your provider.

Once configured, your model provider powers Casibase's chat interface, document Q&A, and other AI features. You can switch between providers anytime to compare results or optimize for cost and performance.

Casibase-providers-view