The production platform for open-weight AI inference
Quick Answer
Together AI's new inference platform enables complete control over open-weight models, optimizing performance and cost while allowing for rapid deployment and iteration.
Quick Take
With support for various model sizes, it significantly reduces startup times, achieving up to 4× faster warm starts for large models like Qwen 3 and Llama 3.3, enhancing user experience and operational efficiency.
Key Points
- Users can deploy models from Together's platform or upload their own open-weight models.
- The platform supports the full lifecycle of models, from fine-tuning to production.
- Startup times for large models are reduced by up to 4× with optimized caching.
- Deployment profiles are pre-optimized to simplify configuration choices for users.
- Control over hardware and optimization profiles allows for tailored performance.
DeepSignal Analysis
What happened
Together AI has launched a new inference platform that allows users to control open-weight models, enhancing performance and reducing costs. The platform reportedly achieves up to 4× faster warm starts for large models, streamlining deployment processes. It supports various model sizes and configurations, making it easier for teams to manage their AI applications.
Key evidence
- Together AI's platform enables users to deploy models from its platform or upload their own, supporting the full lifecycle from fine-tuning to production.
- The platform has been optimized to achieve approximately 4× faster warm starts for large models like Llama 3.3 and Qwen 3.
- Users can implement canary, blue-green, or rolling updates to introduce new model versions gradually, allowing for safer production transitions.
Why it matters
The introduction of this platform addresses the growing demand for control and efficiency in AI model deployment. By allowing teams to manage their models more effectively, it reduces the complexity associated with inference processes. This could lead to faster iteration cycles and improved user experiences, particularly for enterprises that rely on proprietary models and require high performance.
Source Excerpt
Run open models in production with full control over performance, cost, and quality. Deploy in minutes, roll out safely, and scale to your SLOs.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from Together AI
See more →
Open, convenient and predictable: Introducing Provisioned Throughput
Together AI introduces Provisioned Throughput, offering guaranteed inference capacity for MiniMax M3 and GLM-5.2 at $0.05 per PTU per minute, achieving costs up to 90% lower than Claude Opus 4.8. This new model provides predictable pricing and a 99% uptime SLA, catering to companies transitioning to open weight models for production workloads.
.png)