[aiapienthusiast]

Adds a Cheaper Inference provider, mirroring the existing Opper and Requesty providers.
The provider extends OpenAI::ChatProvider and reuses the OpenAI Chat request, response and transform pipeline unchanged.

Cheaper Inference is an OpenAI-compatible LLM gateway.
One API key gives access to models from several labs.

• Base URL: https://api.cheaperinference.com/v1
• API key: CHEAPER_INFERENCE_API_KEY
• Model ids: bare names such as gpt-5.4-mini
• Embeddings: /v1/embeddings, through the same OpenAI::Embedding::RequestType as Opper

It loads through the existing dynamic provider resolution (service: "CheaperInference" to cheaper_inference_provider to CheaperInferenceProvider), so no registry or remap change is needed.

Usage:

cheaper_inference:
  service: "CheaperInference"
  api_key: <%= ENV["CHEAPER_INFERENCE_API_KEY"] %>
  model: "gpt-5.4-mini"

Details

Files:

• lib/active_agent/providers/cheaper_inference_provider.rb
• lib/active_agent/providers/cheaper_inference/options.rb
• lib/active_agent/providers/cheaper_inference/_types.rb
• test/providers/cheaper_inference/provider_loading_test.rb
• test/providers/cheaper_inference/request_test.rb
• README install block and the MCP provider support table in docs/actions/mcps.md, next to the Opper rows

Tests: provider loading and WebMock request tests (chat to /v1/chat/completions, embeddings to /v1/embeddings, CHEAPER_INFERENCE_API_KEY fallback, bare model ids sent unchanged), in the shape of the Opper tests.

Live check: gpt-5.4-mini returned a reply through POST /v1/chat/completions and an embedding model returned a vector through POST /v1/embeddings.