AskWP

Your model,
your choice

Four LLM providers, each a first-class implementation. Switch between them from the admin panel in seconds.

OpenAI OpenAI
Anthropic Anthropic
OpenRouter OpenRouter
Ollama Ollama
01 — OpenAI

OpenAI

Uses the Responses API — the latest generation of OpenAI’s API — with full tool calling support. AskWP sends your conversation history and RAG context to any OpenAI model.

  • Responses API with store: false — conversations never stored on OpenAI’s servers
  • Tool calling — the plugin registers tools the model can invoke during conversation
  • Automatic retries — up to 4 rounds of tool use per turn
  • Token tracking — input and output tokens logged for cost monitoring
responses api 
# What AskWP sends to OpenAI:

POST /v1/responses

{
  "model": "gpt-4o",
  "store": false,
  "input": [
    { "role": "user", "content": "..." }
  ],
  "tools": [
    { "type": "function",
      "name": "open_form" }
  ]
}
02 — Anthropic

Anthropic

Uses the Messages API with native system prompt handling. Claude models receive the system prompt as a dedicated parameter — not embedded in messages — which produces better instruction following.

  • Native system prompt via the system parameter
  • Full tool use support with automatic schema conversion
  • Claude 4 family — Opus, Sonnet, and Haiku
  • Anthropic-version header set automatically
messages api 
# What AskWP sends to Anthropic:

POST /v1/messages

{
  "model": "claude-sonnet-4-5",
  "system": "You are {bot_name}...",
  "messages": [
    { "role": "user", "content": "..." }
  ],
  "tools": [
    { "name": "open_form",
      "input_schema": { ... } }
  ]
}
03 — OpenRouter

OpenRouter

Access any model from any provider through a single API. OpenRouter handles routing, billing, and fallbacks — you just pick a model and go.

  • Any model via a single API key — GPT, Claude, Llama, Mistral, Gemini, and more
  • Unified billing across all providers
  • Automatic fallback routing if a model is unavailable
  • OpenAI-compatible format — works with AskWP’s existing integration
openrouter api 
# What AskWP sends to OpenRouter:

POST /v1/chat/completions

{
  "model": "anthropic/claude-sonnet",
  "messages": [
    { "role": "system", "content": "..." },
    { "role": "user", "content": "..." }
  ],
  "tools": [ ... ]
}

# Any model ID works — switch
# without changing providers.
04 — Ollama

Ollama

Run any open-source model on your own hardware. No API key needed, no per-token costs, complete data privacy. Your conversations never leave your server.

  • OpenAI-compatible endpoint — any model Ollama supports works with AskWP
  • No API key — just point at your Ollama endpoint
  • Zero cost — run as many conversations as your hardware handles
  • Llama, Mistral, Gemma, Phi, DeepSeek, Qwen — your pick
local setup 
# Start Ollama on your server:

$ ollama serve
Listening on 0.0.0.0:11434

# In AskWP settings, enter:

Endpoint: http://localhost:11434
Model:    llama3.1:8b

# That's it. No API key,
# no account, no billing.
05 — Switching

Switch in
seconds

All providers share the same system prompt, RAG context, and conversation format. The plugin handles the translation internally — your visitors see a consistent experience regardless of which model runs behind the scenes.

01

Select

Go to Settings → AskWP → LLM Provider tab. Pick your provider from the dropdown.

02

Configure

Enter your API key (or endpoint for Ollama / OpenRouter). Choose a model name.

03

Save

Hit save. The change takes effect immediately — your next chat message uses the new provider.