What's Happening?
Together AI announced on November 19 the integration of Tencent Hy4 into its serverless chat completions API. This new offering allows developers to utilize Tencent Hy4 with both text and image inputs within the same request, receiving text outputs. The
API supports tool calling and JSON output, featuring a substantial 1,048,576-token context window. Developers can access Hy4 without needing to provision their own infrastructure, simplifying deployment. Together AI provides clear instructions for calling the Hy4 model via an authenticated POST request, setting the model to `tencent/Hy4`, and offers a browser playground for testing. Pricing for the service is set at $0.834 per million tokens for uncached input, $0.042 for cached input, and $2.501 per million tokens for output. The company has demonstrated Hy4's capabilities, including its ability to extract data from image screenshots and recreate charts as HTML files, as well as generating a single-file browser game from a text prompt.
Why It's Important?
The integration of Tencent Hy4 into Together AI's serverless API marks a significant advancement in multimodal AI accessibility for developers. By offering a powerful model that handles both text and image inputs seamlessly, Together AI is lowering the barrier to entry for creating sophisticated AI applications. The large context window of 1,048,576 tokens allows for more complex and nuanced interactions, enabling AI to process extensive information for tasks like detailed content analysis, advanced code generation, and intricate data extraction from visual sources. This development is crucial for industries requiring robust AI solutions for automation, content creation, and data processing, potentially accelerating innovation in areas such as software development, digital media, and business intelligence. The serverless nature of the API also reduces operational overhead for businesses, making advanced AI more cost-effective and scalable.
What's Next?
Following the integration of Tencent Hy4, Together AI is likely to continue expanding its catalog of available models and enhancing its API features. Developers are expected to begin leveraging Hy4's multimodal capabilities for a wide range of applications, from advanced data analytics and automated content generation to interactive AI experiences. The company may introduce further optimizations for performance and cost, potentially adjusting pricing tiers or introducing new usage models based on demand and feedback. Future updates could also include more detailed rate-limit guides and performance benchmarks for Hy4, providing developers with clearer expectations for model behavior under various loads. As more developers adopt Hy4, Together AI will likely gather valuable insights that could inform the development of future AI tools and services, further solidifying its position as a key AI infrastructure provider.
Beyond the Headlines
This development underscores a broader trend in the AI industry towards multimodal models that can understand and process information across different data types, such as text and images. The ability of Tencent Hy4 to interpret visual data and generate structured outputs, like JSON or HTML, highlights the increasing sophistication of AI in bridging the gap between human perception and machine understanding. This could lead to profound shifts in how businesses interact with data, enabling more intuitive and efficient workflows. Ethically, the power of such models to generate complex content from simple prompts raises questions about content authenticity and potential misuse, necessitating ongoing discussions about AI governance and responsible deployment. Legally, the use of AI to generate code or recreate visual content could also bring new challenges related to intellectual property and copyright, as the lines between human and AI-generated work become increasingly blurred.













