Together AI Unveils Cost-Effective On-Demand Dedicated Endpoints

James Ding
Mar 14, 2025 04:21

Together AI introduces Dedicated Endpoints with up to 43% lower pricing, offering enhanced GPU inference capabilities for scaling AI applications, providing high-performance and cost-efficiency.

Together AI has announced the launch of its new on-demand Dedicated Endpoints, designed to offer superior price-performance for GPU inference tasks. This development is aimed at addressing the challenges faced by startups in balancing flexibility and affordability in scaling AI applications, according to Together AI.

Enhanced Performance and Control

The Dedicated Endpoints provide single-tenancy to ensure that user traffic is unaffected by other users, delivering the same high performance as serverless solutions. The offering includes substantial cost savings, full control over deployment hardware and configuration, support for custom fine-tuned models, and no minimum commitments. Users can deploy models such as DeepSeek-R1 and Llama 3.3 70B without incurring upload or storage costs.

Unmatched Cost Savings

With a price reduction of up to 43%, Together AI’s Dedicated Endpoints are positioned as the most cost-effective dedicated GPU inference solution available. The pricing structure offers significant savings compared to other providers, with reductions of up to 50% in some cases. This initiative is part of Together AI’s strategy to provide competitive pricing alongside a broad selection of GPU architectures.

Scalability and Flexibility

Dedicated Endpoints allow businesses to handle usage spikes seamlessly through vertical and horizontal scaling options. Users can scale vertically by increasing GPU count or horizontally by adjusting replica counts to manage peak workloads. This ensures consistent performance and optimized costs, making it suitable for mission-critical AI applications that require reliable QPS and predictable availability.

Deployment Options

Together AI now offers a comprehensive set of deployment options, including serverless, on-demand Dedicated Endpoints, and monthly reserved deployments. Each option provides different benefits, and users can choose based on their specific needs for flexibility, performance, and cost-efficiency. The Dedicated Endpoints are particularly advantageous for customers with strict privacy requirements and those in need of custom model deployment.

In conclusion, Together AI’s Dedicated Endpoints offer a versatile and cost-effective solution for AI companies looking to scale their applications while maintaining high performance and control over their deployments.

Image source: Shutterstock