refactor: move client_type key from provider to model

docs/docs/cli/model.md

@@ -4,7 +4,7 @@ Manage chat and embedding models.
 
 ## model list
 
-List all models with their provider, type, and multimodal flag.
+List all models with their provider, type, client type, and multimodal flag.
 
 ```bash
 memoh model list
@@ -12,7 +12,7 @@ memoh model list
 
 ## model create
 
-Create a new model. Prompts for provider, model ID, type, and (for embedding models) dimensions.
+Create a new model. Prompts for provider, model ID, type, client type, and (for embedding models) dimensions.
 
 ```bash
 memoh model create [options]
@@ -23,6 +23,7 @@ memoh model create [options]
 | `--model_id <id>` | Model ID (e.g. `gpt-4`, `text-embedding-3-small`) |
 | `--name <name>` | Display name |
 | `--provider <provider>` | Provider name |
+| `--client_type <type>` | Client type: `openai-responses`, `openai-completions`, `anthropic-messages`, `google-generative-ai` |
 | `--type <type>` | `chat` or `embedding` |
 | `--dimensions <n>` | Embedding dimensions (required for embedding models) |
 | `--multimodal` | Mark as multimodal |
@@ -30,8 +31,8 @@ memoh model create [options]
 Examples:
 
 ```bash
-memoh model create --model_id gpt-4 --provider my-openai --type chat
-memoh model create --model_id text-embedding-3-small --provider my-openai --type embedding --dimensions 1536
+memoh model create --model_id gpt-4 --provider my-openai --client_type openai-responses --type chat
+memoh model create --model_id text-embedding-3-small --provider my-openai --client_type openai-completions --type embedding --dimensions 1536
 memoh model create
 # Interactive prompts
 ```

docs/docs/cli/provider.md

@@ -1,6 +1,6 @@
 # Provider Commands
 
-Manage LLM providers (OpenAI, Anthropic, Ollama, etc.).
+Manage LLM providers (API endpoints and credentials).
 
 ## provider list
 
@@ -32,16 +32,13 @@ memoh provider create [options]
 | Option | Description |
 |--------|-------------|
 | `--name <name>` | Provider name |
-| `--type <type>` | Client type |
 | `--base_url <url>` | Base URL for the API |
 | `--api_key <key>` | API key |
 
-Supported client types: `openai`, `openai-compat`, `anthropic`, `google`, `azure`, `bedrock`, `mistral`, `xai`, `ollama`, `dashscope`
-
 Examples:
 
 ```bash
-memoh provider create --name my-ollama --type ollama --base_url http://localhost:11434
+memoh provider create --name my-ollama --base_url http://localhost:11434/v1
 memoh provider create
 # Interactive prompts
 ```

docs/docs/concepts/provider-and-model.md

@@ -2,8 +2,19 @@
 
 In Memoh, **provider** and **model** are separate but connected concepts:
 
-- A **provider** is the LLM service configuration (API endpoint, key, client type)
-- A **model** is the concrete chat or embedding model under that provider
+- A **provider** is the LLM service configuration (API endpoint and key)
+- A **model** is the concrete chat or embedding model under that provider, including its **client type** which determines which API protocol to use
+
+## Client Types
+
+Each model has a `client_type` that determines how Memoh communicates with the LLM service:
+
+| Client Type | Description |
+|-------------|-------------|
+| `openai-responses` | OpenAI Responses API |
+| `openai-completions` | OpenAI Chat Completions API (also works with compatible services like Ollama, Mistral, etc.) |
+| `anthropic-messages` | Anthropic Messages API |
+| `google-generative-ai` | Google Generative AI API |
 
 ## Typical Setup
 
@@ -26,4 +37,3 @@ This enables per-bot customization (for quality, latency, or cost).
 
 - `Models > Add Provider > Select Provider > Add Model`
 - `Bots > Select a bot > Settings > Choose chat/memory/embedding models`
-

docs/docs/getting-started.md

@@ -46,10 +46,10 @@ Bots come with a rich set of built-in tools:
 
 ### Multi-LLM Provider Support
 
-Flexibly switch between a wide range of LLM providers:
+Flexibly switch between a wide range of LLM providers via four client types:
 
-- OpenAI, Anthropic, Google, Azure, AWS Bedrock
-- Mistral, XAI, Ollama, Dashscope (Qwen), and more
+- OpenAI Responses API, OpenAI Chat Completions API (including compatible services)
+- Anthropic Messages API, Google Generative AI API
 
 ### MCP Protocol Support
 

docs/docs/getting-started/provider-and-model.md

@@ -14,7 +14,7 @@ Click **Models** in the left sidebar to open the Provider and Model configuratio
 ![Models page - sidebar](/getting-started/provider-model-01-sidebar.png)
 
 The page has two panels:
-- **Left**: Provider list and filter
+- **Left**: Provider list and search
 - **Right**: Selected provider's details and models (or an empty state if none selected)
 
 ## Step 2: Add a Provider
@@ -30,7 +30,6 @@ In the dialog, fill in:
 | **Name** | A display name for this provider (e.g. `my-openai`, `ollama-local`) |
 | **API Key** | Your API key. For local services like Ollama, you can use a placeholder like `ollama` |
 | **Base URL** | The API base URL (e.g. `https://api.openai.com/v1`, `http://localhost:11434/v1` for Ollama) |
-| **Type** | Client type: `openai`, `openai-compat`, `anthropic`, `google`, `azure`, `bedrock`, `mistral`, `xai`, `ollama`, `dashscope` |
 
 ![Add Provider dialog](/getting-started/provider-model-03-provider-dialog.png)
 
@@ -38,13 +37,11 @@ In the dialog, fill in:
 - Name: `openai`
 - API Key: `sk-...`
 - Base URL: `https://api.openai.com/v1`
-- Type: `openai`
 
 **Example — Ollama (local):**
 - Name: `ollama`
 - API Key: `ollama`
 - Base URL: `http://localhost:11434/v1`
-- Type: `ollama`
 
 Click **Add** to save. The new provider appears in the left sidebar.
 
@@ -60,6 +57,7 @@ Fill in:
 
 | Field | Description |
 |-------|-------------|
+| **Client Type** | API protocol: `openai-responses`, `openai-completions`, `anthropic-messages`, or `google-generative-ai` |
 | **Type** | `chat` or `embedding` |
 | **Model** | Model ID (e.g. `gpt-4`, `llama3.2`, `text-embedding-3-small`) |
 | **Display Name** | Optional display name |
@@ -71,8 +69,8 @@ Fill in:
 - One **embedding** model (for memory)
 
 Add them under the same or different providers. For example:
-- Chat: `gpt-4` (OpenAI) or `llama3.2` (Ollama)
-- Embedding: `text-embedding-3-small` (OpenAI) or `nomic-embed-text` (Ollama)
+- Chat: `gpt-4` with client type `openai-responses` (OpenAI) or `llama3.2` with client type `openai-completions` (Ollama)
+- Embedding: `text-embedding-3-small` with client type `openai-completions` (OpenAI) or `nomic-embed-text` with client type `openai-completions` (Ollama)
 
 ## Step 4: Edit or Delete