/ 
Langertha-0.503
/ Langertha::Engine::OpenAI

NAME

Langertha::Engine::OpenAI - OpenAI API

VERSION

version 0.503

SYNOPSIS

use Langertha::Engine::OpenAI;

my $openai = Langertha::Engine::OpenAI->new(
    api_key      => $ENV{OPENAI_API_KEY},
    model        => 'gpt-5.6-terra',
    system_prompt => 'You are a helpful assistant',
    temperature  => 0.7,
);

my $response = $openai->simple_chat('Say something nice');
print $response;

# Embeddings
my $vector = $openai->simple_embedding('Some text to embed');

# Transcription (Whisper)
my $text = $openai->simple_transcription('/path/to/audio.mp3');

# Async with Future::AsyncAwait
use Future::AsyncAwait;

async sub ask_gpt {
    my $response = await $openai->simple_chat_f('What is Perl?');
    say $response;
    my $vector = await $openai->simple_embedding_f('Some text to embed');
}

DESCRIPTION

Provides access to OpenAI's APIs, including GPT models, embeddings, and Whisper transcription. Composes Langertha::Role::OpenAICompatible for the standard OpenAI API format.

Popular models: gpt-6-astra (GPT-6 flagship — 1.05M-token context, 128K max output, text+image input, knowledge cutoff 2026-04-30), gpt-5.6-terra (default, balances intelligence and cost — the GPT-5.6 successor of the former mini tier), gpt-5.6 (Sol, frontier), gpt-5.6-luna (cost-sensitive, successor of the former nano tier), text-embedding-3-large (embeddings), gpt-transcribe (transcription, default), gpt-image-2 (image generation, default).

Dynamic model listing is supported via "list_models" in Langertha::Role::Models. Results are cached for models_cache_ttl seconds (default: 3600).

Get your API key at https://platform.openai.com/ and set LANGERTHA_OPENAI_API_KEY in your environment.

THIS API IS WORK IN PROGRESS

compatibility_for_engine

Optional identifier of the engine this instance is acting as a compatibility shim for. Used internally when one engine is accessed via another's OpenAI endpoint.

whisper

Lazy-built Langertha::Engine::TranscriptionBase instance bound to this engine's api_key, url, transcription_model (gpt-transcribe unless set), user_agent_agent, user_agent_timeout and "connect_address" in Langertha::Role::HTTP, so it transcribes exactly as $openai->simple_transcription does. Useful when you have an OpenAI engine handy and want a focused transcription handle without re-stating credentials:

my $text = $openai->whisper->simple_transcription('/path/audio.mp3');

The transcription engine is fully independent: configure it directly (temperature, language, etc.) via the returned object, or construct your own Langertha::Engine::TranscriptionBase if you need a different model or endpoint.

SEE ALSO

SUPPORT

Issues

Please report bugs and feature requests on GitHub at https://github.com/Getty/langertha/issues.

IRC

Join #langertha on irc.perl.org or message Getty directly.

CONTRIBUTING

Contributions are welcome! Please fork the repository and submit a pull request.

AUTHOR

Torsten Raudssus <getty@cpan.org>

COPYRIGHT AND LICENSE

This software is copyright (c) 2026 by Torsten Raudssus https://raudssus.de/.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.