Changelog

Unreleased

v1.1.1

• Fix a bug when ChatCompletionsCompatible adapter would not work correctly with providers that don't return refusal field (Mistral)
• Fix rare bug when InstructorLite.ask/2 would fail in non-release environments

v1.1.0

• Library now requires Elixir 1.15+
• Introduces new InstructorLite.ask/2 function for cases where you need plain non-structured output.
• To accommodate non-structured output, a new optional InstructorLite.Adapter.find_output/2 callback was added to the adapter behaviour. If you're using a custom adapter, consider implementing it if you want to use InstructorLite.ask/2

v1.0.0

• OpenAI adapter is changed to use the Responses API. The chat completions-based adapter is still available as InstructorLite.Adapters.ChatCompletionsCompatible
• Jason dependency is now optional
• All adapters have been tested and verified for compatibility with reasoning models
• Default prompts have been refined and standardized across adapters

Migrating to 1.0.0

1. If you use the OpenAI adapter and want to continue using the chat completions endpoint, switch to the ChatCompletionsCompatible adapter:

   InstructorLite.instruct(%{
    messages: [
      %{role: "user", content: "John Doe is fourty two years old"}
    ]
   },
   response_model: %{name: :string, age: :integer},
   -   adapter: InstructorLite.Adapters.OpenAI,
   +   adapter: InstructorLite.Adapters.ChatCompletionsCompatible,
   adapter_context: [api_key: Application.fetch_env!(:instructor_lite, :openai_key)]
   )

2. If you use the OpenAI adapter and want to switch to the responses endpoint, update params to comply with the POST /v1/responses interface. Most notably, the messages key should be input:

   InstructorLite.instruct(%{
   -    messages: [
   +    input: [
      %{role: "user", content: "John Doe is fourty two years old"}
    ]
   },
   response_model: %{name: :string, age: :integer},
   adapter: InstructorLite.Adapters.OpenAI,
   adapter_context: [api_key: Application.fetch_env!(:instructor_lite, :openai_key)]
   )

v0.3.0

• Add Gemini adapter

v0.2.0

• [OpenAI] Do not overwrite response_format params key if provided by the user
• Fix consume_response/3 enforcing all keys to present for ad-hoc Ecto schemas
• Add Local Development Guide

v0.1.0

• Initial release