Version Release

Release Note 2025-01-14

1. Fish Speech 1.5 and CosyVoice 2.0 launched: Voice processing upgraded again!

2. Bolt.new integration is live, the ultimate front-end code generation tool.

3. Now supports tablets and large screens, delivering a smoother multi-device experience.

4. The inference.top homepage gets a brand-new look.

5. All-new night mode for an improved nighttime experience.

Release Note 2024-11-22

1. Support for model Qwen 2.5 Coder 32B.

2. BrillAI integrates SD WebUI—welcome to try it out!

3. Starting today, the concurrency limit for free BrillAI users has been increased from 1 to 3.

4. Enhanced UI experience: adjustments made to multi-modal image preview.

Release Note 2024-11-04

1. The audio playback of BrillAI Chat feature has been changed to a streaming interface.

2. Fixed the problem of Credits on the BrillAI recharge page.

Release Note 2024-10-28

1. BrillAI's large language model (LLM) and multimodal (Vision) support voice input, and the generated output responses can be played back in audio form.

2. Bug fix for the mind map sharing feature in BrillAI.

Release Note 2024-10-21

1. Added links to social media platforms on the official website for easy user engagement.

2. The promotion offering 1000 points for new registered and logged-in users ends today. Starting from October 21, 2024, new registered users will receive 500 points.

3. Starting from October 16, 2024, the initial rate limits for paid users' services have been increased from 18 RPM to 3000 RPM, and concurrency has been upgraded from 3 to 1000.

4. The backend has undergone several bug fixes, including fixing errors in the qwen-vl model.

5. The BrillAI login page has been improved.

Release Note 2024-10-14

1. WeChat Pay functionality is now available, allowing for fewer concurrency limits after recharging.

2. Text conversation scrolling has been optimized.

3. A progress bar has been added for image models.

Release Note 2024-09-27

1. Qwen1.5 32B has been upgraded to Qwen2.5 32B, and Qwen2 72B has been upgraded to Qwen2.5 72B. The older models have been discontinued.

2. The Vision and Embedding pages have added Sharing and View Code functionality.

3. The input truncation issue in CosyVoice has been fixed.

Release Note 2024-09-18

1. Support for model Qwen2-vl-7B.

2. Simplified login interface for easier access.

3. Text input now supports LaTeX rendering.

Release Note 2024-09-09

1. CosyVoice-Instruct is now live (supporting onomatopoeia).

2. PayPal integration is now available for recharging, offering fewer concurrency limits after recharging.

3. A new interactive interface for embedding has been added.

Release Note 2024-09-02

1. Supports Qwen2 72B 32K and Internvl2.

2. Added a prompt translation feature for Image scenarios (text-to-image, image-to-image, inpainting) to enhance user interaction experience.

3. Multimodal for text conversation is now available.

4. Added "Like" feature for sharing on the Playground, inviting everyone to interact.

5. A new "Translation Assistant" application has been added to help translate text and speech between multiple languages.

Release Note 2024-08-26

1. New Model Support

• Support for bge-m3, bge-reanker-m3-v2, and SenseVoice;

• The Inpainting model has been upgraded from sd2-inpainting to sd3-medium;

• Support for function calling with Qwen2 72B and Qwen 1.5 32B.

1. Documentation Release

• Documentation (including the API manual) has been released on the inference.top.

1. New Features

• Added a stop button for the text dialogue model;

• Completed mobile adaptation;

• Added the "Run" feature on the Playground page;

• Added parameter explanations on the model-related pages.

Release Note 2024-08-19

1. Voice Scenarios: Add options for on-the-spot recording and sharing functionality.

2. API Code Examples: Include API code examples for Image-to-Image, Inpainting, and audio-related pages.

3. Inpainting: Add features for automatic mask generation and image editing.