v2.29.0•July 24, 2025

Expanded Creative Suite and Enhanced API Reliability

This release marks a major expansion of our creative and multimodal capabilities. We're introducing two highly requested features—Text-to-Speech and Image Editing—for a key provider. Additionally, this update brings significant improvements to API reliability, particularly for streaming responses, and includes a comprehensive update to our model catalog.

New Features

Launched Text-to-Speech (TTS) API: A new audio_speech_processing module enables high-quality audio generation from text for a key provider, including support for raw audio byte responses.
Introduced Image Editing: Added a new image_edit_processing module to support image editing via multipart form data, allowing users to modify images with text prompts on supported models.

Improvements

Enhanced Streaming Reliability: Implemented a fix across multiple providers to ensure a synthetic final 'stop' chunk is always delivered, preventing streams from hanging and improving client-side reliability.
Expanded Model Catalog: Significantly updated the model catalog for a key provider, adding a wide range of new models for chat, image generation, image editing, and text-to-speech.
Added Support for Reasoning Content: API responses from a key provider can now include a reasoning_content field, offering deeper insights into the model's thinking process for complex queries.
Improved Image Generation Handling: Image processing for a major provider now supports downloading images from URLs in addition to base64, and includes more robust error handling for failed generation requests.

Bug Fixes

Corrected Embedding Response Format: Fixed an issue where embedding responses from a provider were missing the required index field, ensuring full compatibility with standard tools.

Chores

Provider Configuration Cleanup: Removed a large number of outdated and deprecated model configuration files to streamline the platform and improve maintainability.