News from Open AI
Try out text-to-speech in the Playground
Building with the text-to-speech API is even easier with the new TTS Playground. You can quickly generate audio by entering a message and selecting from one of six preset voices.
Use function calling in the Playground
With function calling, you can describe functions and have the model intelligently output a JSON object containing arguments to call one or many of your functions. And now you can quickly define and test functions within the Playground.
Generate model outputs that follow your JSON Schemas with Structured Outputs
We just released Structured Outputs, a new feature designed to ensure model-generated outputs will exactly match JSON Schemas that you provide. You can start using Structured Outputs in two forms in the API: through function calling and through the response_format parameter (with a new json_schema option).
Updates
We’ve recently updated our docs page and are regularly adding guides, sample apps, and cookbooks to help you get started and build on the OpenAI platform. Here are a few resources:
- Train, evaluate, and deploy a fine-tuned GPT-4o mini model.
- Call functions with the Chat Completions API.
- Increase reliability with hallucination guardrails.
- Enforce tool use with a customer service example.
- Generate synthetic data for training and fine-tuning models.
- Use CLIP (Contrastive Language–Image Pre-training) embeddings with multimodal RAG.
- Extract and transformdata- Take OCR to new levels
