Announcing Llama 3.3 70B, with enhanced reasoning, mathematics, and instruction-following on Together AI

Together AI has partnered with Meta to support the latest advancement in the Llama model series: Llama-3.3-70B-Instruct 🦙‍

The new Llama 3.3-70B model marks a significant advancement in the AI space, utilizing cutting-edge post-training techniques to elevate its performance across reasoning, mathematics, and instruction-following capabilities. Its performance is comparable to the much larger Llama 3.1 405B model, providing similar capabilities at a fraction of the cost.

More than 250,000 AI developers and enterprises like Zomato, Salesforce, DuckDuckGo, and Washington Post are leveraging the Together AI Inference and Fine-Tuning Platform to innovate with AI and deliver transformative customer experiences.

Serverless Endpoints: Fast, Accurate, and Affordable

To support Llama 3.3-70B, we’re introducing a Together Turbo serverless endpoint that offers an unparalleled combination of performance, accuracy, and affordability. The Llama 3.3 70B serverless endpoint runs FP8, enabling lightning-fast computations without sacrificing the model’s quality.

{{custom-cta-1}}

Dedicated Endpoints: Unmatched Performance and Consistency

For those needing predictable, reliable performance, Together AI will soon offer Dedicated Endpoints for the Llama 3.3-70B model. These endpoints guarantee consistent quality, unaffected by other users' load, making them ideal for mission-critical applications.

Our Dedicated Endpoints come with no rate limits and have high load capacity — they are cost-effective under heavy usage, providing the dedicated resources you need for projects demanding low latency and seamless service. Whether you're building a large-scale application or a highly specialized tool, our dedicated option is here to support your unique needs.

The Growing Momentum of Open Source AI

Llama 3.3 is part of the vibrant ecosystem of open-source models, providing transparency, flexibility, and community-driven innovation. By supporting Llama 3.3-70B, Together AI empowers developers to adapt and enhance these models.

Open-source models mitigate vendor lock-in and give developers full control over their AI infrastructure and their data. Llama 3.3-70B is available under a community license, enabling customization and innovation.

Together AI's commitment to open-source means developers, researchers, and businesses have the freedom to innovate and push boundaries, ensuring powerful AI is accessible to everyone.

Get Started Today

We’re excited to see what you build with Llama 3.3-70B!

Contact our sales team today to learn more about how Together AI and the Llama 3.3-70B model can support your goals.